aspect-based-sentiment-analysis-with-llama-3-1.ipynb

{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.14","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"gpu","dataSources":[{"sourceId":3009440,"sourceType":"datasetVersion","datasetId":1822081},{"sourceId":5968827,"sourceType":"datasetVersion","datasetId":3422393},{"sourceId":9597619,"sourceType":"datasetVersion","datasetId":5783357},{"sourceId":104449,"sourceType":"modelInstanceVersion","modelInstanceId":68809,"modelId":91102},{"sourceId":120005,"sourceType":"modelInstanceVersion","modelInstanceId":100936,"modelId":121027}],"dockerImageVersionId":30762,"isInternetEnabled":true,"language":"python","sourceType":"notebook","isGpuEnabled":true}},"nbformat_minor":4,"nbformat":4,"cells":[{"source":"<a href=\"https://www.kaggle.com/code/yorkyong/aspect-based-sentiment-analysis-with-llama-3-1?scriptVersionId=201631764\" target=\"_blank\"><img align=\"left\" alt=\"Kaggle\" title=\"Open in Kaggle\" src=\"https://kaggle.com/static/images/open-in-kaggle.svg\"></a>","metadata":{},"cell_type":"markdown"},{"cell_type":"code","source":"#with reference from https://www.kaggle.com/code/lucamassaron/fine-tune-llama-2-for-sentiment-analysis","metadata":{"execution":{"iopub.status.busy":"2024-10-17T03:58:02.095147Z","iopub.execute_input":"2024-10-17T03:58:02.095447Z","iopub.status.idle":"2024-10-17T03:58:02.099964Z","shell.execute_reply.started":"2024-10-17T03:58:02.095415Z","shell.execute_reply":"2024-10-17T03:58:02.099207Z"},"trusted":true},"execution_count":1,"outputs":[]},{"cell_type":"code","source":"!pip install -q -U torch --index-url https://download.pytorch.org/whl/cu117\n!pip install -q -U -i https://pypi.org/simple/ bitsandbytes\n!pip install -q -U transformers\n!pip install -q -U accelerate\n!pip install -q -U datasets\n!pip install -q -U trl\n!pip install -q -U peft","metadata":{"execution":{"iopub.status.busy":"2024-10-17T03:58:02.101373Z","iopub.execute_input":"2024-10-17T03:58:02.101711Z","iopub.status.idle":"2024-10-17T03:59:56.942931Z","shell.execute_reply.started":"2024-10-17T03:58:02.101679Z","shell.execute_reply":"2024-10-17T03:59:56.941665Z"},"trusted":true},"execution_count":2,"outputs":[]},{"cell_type":"code","source":"import os\nos.environ[\"CUDA_VISIBLE_DEVICES\"] = \"0\"\nos.environ[\"TOKENIZERS_PARALLELISM\"] = \"true\"","metadata":{"execution":{"iopub.status.busy":"2024-10-17T03:59:56.945201Z","iopub.execute_input":"2024-10-17T03:59:56.945543Z","iopub.status.idle":"2024-10-17T03:59:56.95072Z","shell.execute_reply.started":"2024-10-17T03:59:56.94551Z","shell.execute_reply":"2024-10-17T03:59:56.949853Z"},"trusted":true},"execution_count":3,"outputs":[]},{"cell_type":"code","source":"import warnings\nwarnings.filterwarnings(\"ignore\")","metadata":{"execution":{"iopub.status.busy":"2024-10-17T03:59:56.951699Z","iopub.execute_input":"2024-10-17T03:59:56.951947Z","iopub.status.idle":"2024-10-17T03:59:56.962834Z","shell.execute_reply.started":"2024-10-17T03:59:56.951919Z","shell.execute_reply":"2024-10-17T03:59:56.961967Z"},"trusted":true},"execution_count":4,"outputs":[]},{"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nimport seaborn as sns\nimport datetime as dt\nimport re\nimport os\nfrom tqdm import tqdm\nimport bitsandbytes as bnb\nimport torch\nimport torch.nn as nn\nimport transformers\nfrom datasets import Dataset\nfrom peft import LoraConfig, PeftConfig\nfrom trl import SFTTrainer\nfrom trl import setup_chat_format\nfrom transformers import (AutoConfig,\n                          AutoModelForCausalLM, \n                          AutoTokenizer,\n                          LlamaForCausalLM, \n                          LlamaTokenizer,\n                          BitsAndBytesConfig, \n                          TrainingArguments, \n                          pipeline, \n                          logging)\nfrom sklearn.metrics import (accuracy_score, \n                             classification_report, \n                             confusion_matrix)\nfrom sklearn.model_selection import train_test_split\n","metadata":{"execution":{"iopub.status.busy":"2024-10-17T03:59:56.965256Z","iopub.execute_input":"2024-10-17T03:59:56.965985Z","iopub.status.idle":"2024-10-17T04:00:16.207409Z","shell.execute_reply.started":"2024-10-17T03:59:56.965941Z","shell.execute_reply":"2024-10-17T04:00:16.206465Z"},"trusted":true},"execution_count":5,"outputs":[]},{"cell_type":"code","source":"pd.set_option('display.max_colwidth', None)  # Show full width of each column","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:00:16.208506Z","iopub.execute_input":"2024-10-17T04:00:16.209145Z","iopub.status.idle":"2024-10-17T04:00:16.213786Z","shell.execute_reply.started":"2024-10-17T04:00:16.209101Z","shell.execute_reply":"2024-10-17T04:00:16.212711Z"},"trusted":true},"execution_count":6,"outputs":[]},{"cell_type":"code","source":"print(f\"pytorch version {torch.__version__}\")","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:00:16.215181Z","iopub.execute_input":"2024-10-17T04:00:16.215785Z","iopub.status.idle":"2024-10-17T04:00:16.264885Z","shell.execute_reply.started":"2024-10-17T04:00:16.215741Z","shell.execute_reply":"2024-10-17T04:00:16.263892Z"},"trusted":true},"execution_count":7,"outputs":[{"name":"stdout","text":"pytorch version 2.4.0\n","output_type":"stream"}]},{"cell_type":"code","source":"device = torch.device(\"cuda:0\" if torch.cuda.is_available() else \"cpu\")\nprint(f\"working on {device}\")","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:00:16.266076Z","iopub.execute_input":"2024-10-17T04:00:16.266366Z","iopub.status.idle":"2024-10-17T04:00:16.277615Z","shell.execute_reply.started":"2024-10-17T04:00:16.266334Z","shell.execute_reply":"2024-10-17T04:00:16.276718Z"},"trusted":true},"execution_count":8,"outputs":[{"name":"stdout","text":"working on cuda:0\n","output_type":"stream"}]},{"cell_type":"markdown","source":"# Preparing the data and the core evaluation functions","metadata":{}},{"cell_type":"code","source":"# filename= '/kaggle/input/mcdonalds-review-dataset-6-months-clean-sentiment/cleaned_data _for_ABSA.xlsx'\nfilename= '/kaggle/input/mcdonalds-store-reviews/McDonald_s_Reviews.csv' #To identify use more Neutral sentiments + Aspects = Price for data augmentation","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:00:21.779871Z","iopub.execute_input":"2024-10-17T04:00:21.780759Z","iopub.status.idle":"2024-10-17T04:00:21.784778Z","shell.execute_reply.started":"2024-10-17T04:00:21.780716Z","shell.execute_reply":"2024-10-17T04:00:21.783661Z"},"trusted":true},"execution_count":9,"outputs":[]},{"cell_type":"code","source":"df = pd.read_csv(filename,encoding=\"utf8\", encoding_errors=\"replace\")","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:01:18.51945Z","iopub.execute_input":"2024-10-17T04:01:18.519852Z","iopub.status.idle":"2024-10-17T04:01:18.783307Z","shell.execute_reply.started":"2024-10-17T04:01:18.519817Z","shell.execute_reply":"2024-10-17T04:01:18.782495Z"},"trusted":true},"execution_count":11,"outputs":[]},{"cell_type":"code","source":"df.info()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:01:20.592821Z","iopub.execute_input":"2024-10-17T04:01:20.593204Z","iopub.status.idle":"2024-10-17T04:01:20.643756Z","shell.execute_reply.started":"2024-10-17T04:01:20.593168Z","shell.execute_reply":"2024-10-17T04:01:20.642763Z"},"trusted":true},"execution_count":12,"outputs":[{"name":"stdout","text":"<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 33396 entries, 0 to 33395\nData columns (total 10 columns):\n #   Column         Non-Null Count  Dtype  \n---  ------         --------------  -----  \n 0   reviewer_id    33396 non-null  int64  \n 1   store_name     33396 non-null  object \n 2   category       33396 non-null  object \n 3   store_address  33396 non-null  object \n 4   latitude       32736 non-null  float64\n 5   longitude      32736 non-null  float64\n 6   rating_count   33396 non-null  object \n 7   review_time    33396 non-null  object \n 8   review         33396 non-null  object \n 9   rating         33396 non-null  object \ndtypes: float64(2), int64(1), object(7)\nmemory usage: 2.5+ MB\n","output_type":"stream"}]},{"cell_type":"code","source":"df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:01:24.26635Z","iopub.execute_input":"2024-10-17T04:01:24.266769Z","iopub.status.idle":"2024-10-17T04:01:24.288147Z","shell.execute_reply.started":"2024-10-17T04:01:24.26673Z","shell.execute_reply":"2024-10-17T04:01:24.287266Z"},"trusted":true},"execution_count":13,"outputs":[{"execution_count":13,"output_type":"execute_result","data":{"text/plain":"   reviewer_id  store_name              category  \\\n0            1  McDonald's  Fast food restaurant   \n1            2  McDonald's  Fast food restaurant   \n2            3  McDonald's  Fast food restaurant   \n3            4  McDonald's  Fast food restaurant   \n4            5  McDonald's  Fast food restaurant   \n\n                                       store_address  latitude   longitude  \\\n0  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n1  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n2  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n3  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n4  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n\n  rating_count   review_time  \\\n0        1,240  3 months ago   \n1        1,240    5 days ago   \n2        1,240    5 days ago   \n3        1,240   a month ago   \n4        1,240  2 months ago   \n\n                                                                                                                                                                                                                                                                                                                                                                         review  \\\n0                                                                                                          Why does it look like someone spit on my food?\\nI had a normal transaction,  everyone was chill and polite, but now i dont want to eat this. Im trying not to think about what this milky white/clear substance is all over my food, i d*** sure am not coming back.   \n1                                                                                                                                 It'd McDonalds. It is what it is as far as the food and atmosphere go. The staff here does make a difference. They are all friendly, accommodating and always smiling. Makes for a more pleasant experience than many other fast food places.   \n2  Made a mobile order got to the speaker and checked it in.\\nLine was not moving so I had to leave otherwise I���������������������������d be late for work.\\nNever got the refund in the app.\\nI called them and they said I could only get my money back in person because it was stuck in the system.\\nWent there in person the next day  and the manager told me she wasn�   \n3                                                                                                                                                                                                                                                                      My mc. Crispy chicken sandwich was ������������������������������������ customer service was quick and p   \n4                                                      I repeat my order 3 times in the drive thru, and she still manage to mess it up , it was suppose to be a large meal double filet of fish with large fries , no cheese . It was all wrong , they either need to pay close attention to the order being made , understand English or they need not to work at a drive thru   \n\n    rating  \n0   1 star  \n1  4 stars  \n2   1 star  \n3  5 stars  \n4   1 star  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>reviewer_id</th>\n      <th>store_name</th>\n      <th>category</th>\n      <th>store_address</th>\n      <th>latitude</th>\n      <th>longitude</th>\n      <th>rating_count</th>\n      <th>review_time</th>\n      <th>review</th>\n      <th>rating</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>3 months ago</td>\n      <td>Why does it look like someone spit on my food?\\nI had a normal transaction,  everyone was chill and polite, but now i dont want to eat this. Im trying not to think about what this milky white/clear substance is all over my food, i d*** sure am not coming back.</td>\n      <td>1 star</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>5 days ago</td>\n      <td>It'd McDonalds. It is what it is as far as the food and atmosphere go. The staff here does make a difference. They are all friendly, accommodating and always smiling. Makes for a more pleasant experience than many other fast food places.</td>\n      <td>4 stars</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>5 days ago</td>\n      <td>Made a mobile order got to the speaker and checked it in.\\nLine was not moving so I had to leave otherwise I���������������������������d be late for work.\\nNever got the refund in the app.\\nI called them and they said I could only get my money back in person because it was stuck in the system.\\nWent there in person the next day  and the manager told me she wasn�</td>\n      <td>1 star</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>4</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>a month ago</td>\n      <td>My mc. Crispy chicken sandwich was ������������������������������������ customer service was quick and p</td>\n      <td>5 stars</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>5</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>2 months ago</td>\n      <td>I repeat my order 3 times in the drive thru, and she still manage to mess it up , it was suppose to be a large meal double filet of fish with large fries , no cheese . It was all wrong , they either need to pay close attention to the order being made , understand English or they need not to work at a drive thru</td>\n      <td>1 star</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"df.review_time.unique()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:02:03.507614Z","iopub.execute_input":"2024-10-17T04:02:03.508489Z","iopub.status.idle":"2024-10-17T04:02:03.517316Z","shell.execute_reply.started":"2024-10-17T04:02:03.508451Z","shell.execute_reply":"2024-10-17T04:02:03.516421Z"},"trusted":true},"execution_count":14,"outputs":[{"execution_count":14,"output_type":"execute_result","data":{"text/plain":"array(['3 months ago', '5 days ago', 'a month ago', '2 months ago',\n       '3 weeks ago', 'a year ago', '6 months ago', '5 months ago',\n       '7 months ago', '10 months ago', '4 years ago', '8 months ago',\n       '4 months ago', '3 years ago', '2 years ago', '9 months ago',\n       '11 months ago', '5 years ago', '6 years ago', '7 years ago',\n       '8 years ago', '3 days ago', 'a week ago', '2 days ago',\n       '2 weeks ago', '10 years ago', '9 years ago', '6 days ago',\n       '4 weeks ago', '21 hours ago', '4 days ago', '11 years ago',\n       '12 years ago', 'a day ago', '23 hours ago', '6 hours ago',\n       '20 hours ago', '22 hours ago', '8 hours ago'], dtype=object)"},"metadata":{}}]},{"cell_type":"code","source":"def map_rating_to_sentiment(rating):\n    if rating in ['4 stars', '5 stars']:\n        return 'positive'\n    elif rating in ['1 star', '2 stars']:\n        return 'negative'\n    else:\n        return 'neutral'\n\n# Assuming your DataFrame is named 'df' and the ratings column is named 'ratings'\ndf['sentiment'] = df['rating'].apply(map_rating_to_sentiment)","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:07:19.708494Z","iopub.execute_input":"2024-10-17T04:07:19.709207Z","iopub.status.idle":"2024-10-17T04:07:19.72717Z","shell.execute_reply.started":"2024-10-17T04:07:19.70917Z","shell.execute_reply":"2024-10-17T04:07:19.726245Z"},"trusted":true},"execution_count":15,"outputs":[]},{"cell_type":"code","source":"def convert_to_datetime(time_str):\n    current_time = dt.datetime.now()\n    if 'year' in time_str:\n        years = int(re.findall(r'\\d+', time_str)[0]) if re.search(r'\\d+', time_str) else 1\n        return current_time - dt.timedelta(days=years*365)\n    elif 'month' in time_str:\n        months = int(re.findall(r'\\d+', time_str)[0]) if re.search(r'\\d+', time_str) else 1\n        return current_time - dt.timedelta(days=months*30)\n    elif 'week' in time_str:\n        weeks = int(re.findall(r'\\d+', time_str)[0]) if re.search(r'\\d+', time_str) else 1\n        return current_time - dt.timedelta(weeks=weeks)\n    elif 'day' in time_str:\n        days = int(re.findall(r'\\d+', time_str)[0]) if re.search(r'\\d+', time_str) else 1\n        return current_time - dt.timedelta(days=days)\n    elif 'hour' in time_str:\n        hours = int(re.findall(r'\\d+', time_str)[0]) if re.search(r'\\d+', time_str) else 1\n        return current_time - dt.timedelta(hours=hours)\n    else:\n        return current_time\n\n# Apply the function to create a new column with actual dates\ndf['actual_date'] = df['review_time'].apply(convert_to_datetime)\n","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:07:20.526423Z","iopub.execute_input":"2024-10-17T04:07:20.527264Z","iopub.status.idle":"2024-10-17T04:07:20.761092Z","shell.execute_reply.started":"2024-10-17T04:07:20.527224Z","shell.execute_reply":"2024-10-17T04:07:20.760086Z"},"trusted":true},"execution_count":16,"outputs":[]},{"cell_type":"code","source":"# Drop unknown and missing rows\n\ndf['review'] = df['review'].apply(lambda x: re.sub(r'[^a-zA-Z0-9\\s]+', '', x) if isinstance(x, str) else x).copy()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:07:21.627185Z","iopub.execute_input":"2024-10-17T04:07:21.627567Z","iopub.status.idle":"2024-10-17T04:07:21.870808Z","shell.execute_reply.started":"2024-10-17T04:07:21.627528Z","shell.execute_reply":"2024-10-17T04:07:21.869973Z"},"trusted":true},"execution_count":17,"outputs":[]},{"cell_type":"code","source":"df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:07:28.814154Z","iopub.execute_input":"2024-10-17T04:07:28.81508Z","iopub.status.idle":"2024-10-17T04:07:28.831867Z","shell.execute_reply.started":"2024-10-17T04:07:28.815042Z","shell.execute_reply":"2024-10-17T04:07:28.830847Z"},"trusted":true},"execution_count":19,"outputs":[{"execution_count":19,"output_type":"execute_result","data":{"text/plain":"   reviewer_id  store_name              category  \\\n0            1  McDonald's  Fast food restaurant   \n1            2  McDonald's  Fast food restaurant   \n2            3  McDonald's  Fast food restaurant   \n3            4  McDonald's  Fast food restaurant   \n4            5  McDonald's  Fast food restaurant   \n\n                                       store_address  latitude   longitude  \\\n0  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n1  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n2  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n3  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n4  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n\n  rating_count   review_time  \\\n0        1,240  3 months ago   \n1        1,240    5 days ago   \n2        1,240    5 days ago   \n3        1,240   a month ago   \n4        1,240  2 months ago   \n\n                                                                                                                                                                                                                                                                                                                                         review  \\\n0                                                                                    Why does it look like someone spit on my food\\nI had a normal transaction  everyone was chill and polite but now i dont want to eat this Im trying not to think about what this milky whiteclear substance is all over my food i d sure am not coming back   \n1                                                                                                        Itd McDonalds It is what it is as far as the food and atmosphere go The staff here does make a difference They are all friendly accommodating and always smiling Makes for a more pleasant experience than many other fast food places   \n2  Made a mobile order got to the speaker and checked it in\\nLine was not moving so I had to leave otherwise Id be late for work\\nNever got the refund in the app\\nI called them and they said I could only get my money back in person because it was stuck in the system\\nWent there in person the next day  and the manager told me she wasn   \n3                                                                                                                                                                                                                                                                           My mc Crispy chicken sandwich was  customer service was quick and p   \n4                            I repeat my order 3 times in the drive thru and she still manage to mess it up  it was suppose to be a large meal double filet of fish with large fries  no cheese  It was all wrong  they either need to pay close attention to the order being made  understand English or they need not to work at a drive thru   \n\n    rating sentiment                actual_date  \n0   1 star  negative 2024-07-19 04:07:20.534412  \n1  4 stars  positive 2024-10-12 04:07:20.534628  \n2   1 star  negative 2024-10-12 04:07:20.534639  \n3  5 stars  positive 2024-09-17 04:07:20.534647  \n4   1 star  negative 2024-08-18 04:07:20.534653  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>reviewer_id</th>\n      <th>store_name</th>\n      <th>category</th>\n      <th>store_address</th>\n      <th>latitude</th>\n      <th>longitude</th>\n      <th>rating_count</th>\n      <th>review_time</th>\n      <th>review</th>\n      <th>rating</th>\n      <th>sentiment</th>\n      <th>actual_date</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>1</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>3 months ago</td>\n      <td>Why does it look like someone spit on my food\\nI had a normal transaction  everyone was chill and polite but now i dont want to eat this Im trying not to think about what this milky whiteclear substance is all over my food i d sure am not coming back</td>\n      <td>1 star</td>\n      <td>negative</td>\n      <td>2024-07-19 04:07:20.534412</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>2</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>5 days ago</td>\n      <td>Itd McDonalds It is what it is as far as the food and atmosphere go The staff here does make a difference They are all friendly accommodating and always smiling Makes for a more pleasant experience than many other fast food places</td>\n      <td>4 stars</td>\n      <td>positive</td>\n      <td>2024-10-12 04:07:20.534628</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>3</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>5 days ago</td>\n      <td>Made a mobile order got to the speaker and checked it in\\nLine was not moving so I had to leave otherwise Id be late for work\\nNever got the refund in the app\\nI called them and they said I could only get my money back in person because it was stuck in the system\\nWent there in person the next day  and the manager told me she wasn</td>\n      <td>1 star</td>\n      <td>negative</td>\n      <td>2024-10-12 04:07:20.534639</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>4</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>a month ago</td>\n      <td>My mc Crispy chicken sandwich was  customer service was quick and p</td>\n      <td>5 stars</td>\n      <td>positive</td>\n      <td>2024-09-17 04:07:20.534647</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>5</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>2 months ago</td>\n      <td>I repeat my order 3 times in the drive thru and she still manage to mess it up  it was suppose to be a large meal double filet of fish with large fries  no cheese  It was all wrong  they either need to pay close attention to the order being made  understand English or they need not to work at a drive thru</td>\n      <td>1 star</td>\n      <td>negative</td>\n      <td>2024-08-18 04:07:20.534653</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"print(f\"The dataframe for modelling has {len(df)} rows.\")","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:07:47.007451Z","iopub.execute_input":"2024-10-17T04:07:47.00786Z","iopub.status.idle":"2024-10-17T04:07:47.012925Z","shell.execute_reply.started":"2024-10-17T04:07:47.007822Z","shell.execute_reply":"2024-10-17T04:07:47.011968Z"},"trusted":true},"execution_count":20,"outputs":[{"name":"stdout","text":"The dataframe for modelling has 33396 rows.\n","output_type":"stream"}]},{"cell_type":"code","source":"# Group by review_time and sentiment, then compute the size of each group\nsentiment_counts = df.groupby(['review_time', 'sentiment']).size().unstack(fill_value=0)\n\nsentiment_counts","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:09:04.20467Z","iopub.execute_input":"2024-10-17T04:09:04.205549Z","iopub.status.idle":"2024-10-17T04:09:04.235657Z","shell.execute_reply.started":"2024-10-17T04:09:04.205506Z","shell.execute_reply":"2024-10-17T04:09:04.234666Z"},"trusted":true},"execution_count":21,"outputs":[{"execution_count":21,"output_type":"execute_result","data":{"text/plain":"sentiment      negative  neutral  positive\nreview_time                               \n10 months ago       216       73       214\n10 years ago         11       15        12\n11 months ago       201       48       208\n11 years ago          7        1         2\n12 years ago          0        1         3\n2 days ago            9        3        20\n2 months ago        272       77       276\n2 weeks ago          68       17        52\n2 years ago        1448      498      1946\n20 hours ago          1        0         0\n21 hours ago          0        0         2\n22 hours ago          1        0         0\n23 hours ago          1        0         0\n3 days ago           13        2         7\n3 months ago        227       41       224\n3 weeks ago          46       12        64\n3 years ago        1783      842      2897\n4 days ago           10        8         8\n4 months ago        147       35       211\n4 weeks ago          21        3        23\n4 years ago        2299     1045      3396\n5 days ago           11        5         7\n5 months ago        181       58       219\n5 years ago        1681      684      1941\n6 days ago            9        2         6\n6 hours ago           1        0         0\n6 months ago        166       66       224\n6 years ago         610      318       751\n7 months ago        179       72       221\n7 years ago         217       47       123\n8 hours ago           1        0         0\n8 months ago        234       47       217\n8 years ago          61       10        20\n9 months ago        162       65       217\n9 years ago          33        8        11\na day ago             9        0         8\na month ago         198       63       232\na week ago           57       19        49\na year ago         1926      633      2250","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th>sentiment</th>\n      <th>negative</th>\n      <th>neutral</th>\n      <th>positive</th>\n    </tr>\n    <tr>\n      <th>review_time</th>\n      <th></th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>10 months ago</th>\n      <td>216</td>\n      <td>73</td>\n      <td>214</td>\n    </tr>\n    <tr>\n      <th>10 years ago</th>\n      <td>11</td>\n      <td>15</td>\n      <td>12</td>\n    </tr>\n    <tr>\n      <th>11 months ago</th>\n      <td>201</td>\n      <td>48</td>\n      <td>208</td>\n    </tr>\n    <tr>\n      <th>11 years ago</th>\n      <td>7</td>\n      <td>1</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>12 years ago</th>\n      <td>0</td>\n      <td>1</td>\n      <td>3</td>\n    </tr>\n    <tr>\n      <th>2 days ago</th>\n      <td>9</td>\n      <td>3</td>\n      <td>20</td>\n    </tr>\n    <tr>\n      <th>2 months ago</th>\n      <td>272</td>\n      <td>77</td>\n      <td>276</td>\n    </tr>\n    <tr>\n      <th>2 weeks ago</th>\n      <td>68</td>\n      <td>17</td>\n      <td>52</td>\n    </tr>\n    <tr>\n      <th>2 years ago</th>\n      <td>1448</td>\n      <td>498</td>\n      <td>1946</td>\n    </tr>\n    <tr>\n      <th>20 hours ago</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>21 hours ago</th>\n      <td>0</td>\n      <td>0</td>\n      <td>2</td>\n    </tr>\n    <tr>\n      <th>22 hours ago</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>23 hours ago</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>3 days ago</th>\n      <td>13</td>\n      <td>2</td>\n      <td>7</td>\n    </tr>\n    <tr>\n      <th>3 months ago</th>\n      <td>227</td>\n      <td>41</td>\n      <td>224</td>\n    </tr>\n    <tr>\n      <th>3 weeks ago</th>\n      <td>46</td>\n      <td>12</td>\n      <td>64</td>\n    </tr>\n    <tr>\n      <th>3 years ago</th>\n      <td>1783</td>\n      <td>842</td>\n      <td>2897</td>\n    </tr>\n    <tr>\n      <th>4 days ago</th>\n      <td>10</td>\n      <td>8</td>\n      <td>8</td>\n    </tr>\n    <tr>\n      <th>4 months ago</th>\n      <td>147</td>\n      <td>35</td>\n      <td>211</td>\n    </tr>\n    <tr>\n      <th>4 weeks ago</th>\n      <td>21</td>\n      <td>3</td>\n      <td>23</td>\n    </tr>\n    <tr>\n      <th>4 years ago</th>\n      <td>2299</td>\n      <td>1045</td>\n      <td>3396</td>\n    </tr>\n    <tr>\n      <th>5 days ago</th>\n      <td>11</td>\n      <td>5</td>\n      <td>7</td>\n    </tr>\n    <tr>\n      <th>5 months ago</th>\n      <td>181</td>\n      <td>58</td>\n      <td>219</td>\n    </tr>\n    <tr>\n      <th>5 years ago</th>\n      <td>1681</td>\n      <td>684</td>\n      <td>1941</td>\n    </tr>\n    <tr>\n      <th>6 days ago</th>\n      <td>9</td>\n      <td>2</td>\n      <td>6</td>\n    </tr>\n    <tr>\n      <th>6 hours ago</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>6 months ago</th>\n      <td>166</td>\n      <td>66</td>\n      <td>224</td>\n    </tr>\n    <tr>\n      <th>6 years ago</th>\n      <td>610</td>\n      <td>318</td>\n      <td>751</td>\n    </tr>\n    <tr>\n      <th>7 months ago</th>\n      <td>179</td>\n      <td>72</td>\n      <td>221</td>\n    </tr>\n    <tr>\n      <th>7 years ago</th>\n      <td>217</td>\n      <td>47</td>\n      <td>123</td>\n    </tr>\n    <tr>\n      <th>8 hours ago</th>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>8 months ago</th>\n      <td>234</td>\n      <td>47</td>\n      <td>217</td>\n    </tr>\n    <tr>\n      <th>8 years ago</th>\n      <td>61</td>\n      <td>10</td>\n      <td>20</td>\n    </tr>\n    <tr>\n      <th>9 months ago</th>\n      <td>162</td>\n      <td>65</td>\n      <td>217</td>\n    </tr>\n    <tr>\n      <th>9 years ago</th>\n      <td>33</td>\n      <td>8</td>\n      <td>11</td>\n    </tr>\n    <tr>\n      <th>a day ago</th>\n      <td>9</td>\n      <td>0</td>\n      <td>8</td>\n    </tr>\n    <tr>\n      <th>a month ago</th>\n      <td>198</td>\n      <td>63</td>\n      <td>232</td>\n    </tr>\n    <tr>\n      <th>a week ago</th>\n      <td>57</td>\n      <td>19</td>\n      <td>49</td>\n    </tr>\n    <tr>\n      <th>a year ago</th>\n      <td>1926</td>\n      <td>633</td>\n      <td>2250</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"# Define the relevant time periods for filtering\ntime_periods = ['11 months ago', '10 months ago', '9 months ago', '8 months ago', '7 months ago']\n\n# Filter the dataframe based on review_time\nfiltered_df = df[df['review_time'].isin(time_periods)]\n\nfiltered_df.info()\n","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:47.496872Z","iopub.execute_input":"2024-10-17T04:18:47.497298Z","iopub.status.idle":"2024-10-17T04:18:47.516995Z","shell.execute_reply.started":"2024-10-17T04:18:47.497261Z","shell.execute_reply":"2024-10-17T04:18:47.516051Z"},"trusted":true},"execution_count":38,"outputs":[{"name":"stdout","text":"<class 'pandas.core.frame.DataFrame'>\nIndex: 2374 entries, 18 to 33380\nData columns (total 12 columns):\n #   Column         Non-Null Count  Dtype         \n---  ------         --------------  -----         \n 0   reviewer_id    2374 non-null   int64         \n 1   store_name     2374 non-null   object        \n 2   category       2374 non-null   object        \n 3   store_address  2374 non-null   object        \n 4   latitude       2309 non-null   float64       \n 5   longitude      2309 non-null   float64       \n 6   rating_count   2374 non-null   object        \n 7   review_time    2374 non-null   object        \n 8   review         2374 non-null   object        \n 9   rating         2374 non-null   object        \n 10  sentiment      2374 non-null   object        \n 11  actual_date    2374 non-null   datetime64[ns]\ndtypes: datetime64[ns](1), float64(2), int64(1), object(8)\nmemory usage: 241.1+ KB\n","output_type":"stream"}]},{"cell_type":"code","source":"# Group by review_time and sentiment, then compute the size of each group\nsentiment_counts1 = filtered_df.groupby(['review_time', 'sentiment']).size().unstack(fill_value=0)\n\nsentiment_counts1","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:48.236451Z","iopub.execute_input":"2024-10-17T04:18:48.236854Z","iopub.status.idle":"2024-10-17T04:18:48.250319Z","shell.execute_reply.started":"2024-10-17T04:18:48.236815Z","shell.execute_reply":"2024-10-17T04:18:48.249389Z"},"trusted":true},"execution_count":39,"outputs":[{"execution_count":39,"output_type":"execute_result","data":{"text/plain":"sentiment      negative  neutral  positive\nreview_time                               \n10 months ago       216       73       214\n11 months ago       201       48       208\n7 months ago        179       72       221\n8 months ago        234       47       217\n9 months ago        162       65       217","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th>sentiment</th>\n      <th>negative</th>\n      <th>neutral</th>\n      <th>positive</th>\n    </tr>\n    <tr>\n      <th>review_time</th>\n      <th></th>\n      <th></th>\n      <th></th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>10 months ago</th>\n      <td>216</td>\n      <td>73</td>\n      <td>214</td>\n    </tr>\n    <tr>\n      <th>11 months ago</th>\n      <td>201</td>\n      <td>48</td>\n      <td>208</td>\n    </tr>\n    <tr>\n      <th>7 months ago</th>\n      <td>179</td>\n      <td>72</td>\n      <td>221</td>\n    </tr>\n    <tr>\n      <th>8 months ago</th>\n      <td>234</td>\n      <td>47</td>\n      <td>217</td>\n    </tr>\n    <tr>\n      <th>9 months ago</th>\n      <td>162</td>\n      <td>65</td>\n      <td>217</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"markdown","source":"# Prompt","metadata":{}},{"cell_type":"code","source":"def generate_prompt(review):\n    return f\"\"\"\nAnalyze the sentiment of the following review for five specific aspects: service, cleanliness, food, price, and others.\nFor each aspect, assign a sentiment score based on the following scale:\n0: Aspect is not mentioned\n1: Negative sentiment\n2: Neutral sentiment\n3: Positive sentiment\nIMPORTANT:\n- If the review is too brief or generic and not about specific aspects, assess the overall sentiment and assign it to others.\n- Only provide scores for aspects that are EXPLICITLY mentioned or can be DIRECTLY inferred from the review.\n- If an aspect is not mentioned or cannot be directly inferred, assign it a score of 0.\n- Use the 'others' category to capture any significant aspects mentioned that don't fit into service, cleanliness, food, or price.\n- Do NOT repeat the review or provide any additional analysis or explanations beyond the scores.\n- Provide the scores ONLY ONCE in the exact format shown below.\nProvide your analysis in the following format ONLY:\nservice=[score]\ncleanliness=[score]\nfood=[score]\nprice=[score]\nothers=[score]\nReview: {review}\nAnalysis:\n\"\"\"","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:50.254065Z","iopub.execute_input":"2024-10-17T04:18:50.25451Z","iopub.status.idle":"2024-10-17T04:18:50.260404Z","shell.execute_reply.started":"2024-10-17T04:18:50.254468Z","shell.execute_reply":"2024-10-17T04:18:50.259324Z"},"trusted":true},"execution_count":40,"outputs":[]},{"cell_type":"code","source":"# # Generate the full prompt (input) and retain sentiment as a separate column\n# df['prompt'] = df.apply(generate_prompt, axis=1)\n\n# Now convert the DataFrame to a dataset\ndf_data = Dataset.from_pandas(filtered_df)","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:51.005854Z","iopub.execute_input":"2024-10-17T04:18:51.006247Z","iopub.status.idle":"2024-10-17T04:18:51.037102Z","shell.execute_reply.started":"2024-10-17T04:18:51.006208Z","shell.execute_reply":"2024-10-17T04:18:51.036333Z"},"trusted":true},"execution_count":41,"outputs":[]},{"cell_type":"code","source":"df_data","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:51.743236Z","iopub.execute_input":"2024-10-17T04:18:51.744049Z","iopub.status.idle":"2024-10-17T04:18:51.750116Z","shell.execute_reply.started":"2024-10-17T04:18:51.744008Z","shell.execute_reply":"2024-10-17T04:18:51.74889Z"},"trusted":true},"execution_count":42,"outputs":[{"execution_count":42,"output_type":"execute_result","data":{"text/plain":"Dataset({\n    features: ['reviewer_id', 'store_name', 'category', 'store_address', 'latitude ', 'longitude', 'rating_count', 'review_time', 'review', 'rating', 'sentiment', 'actual_date', '__index_level_0__'],\n    num_rows: 2374\n})"},"metadata":{}}]},{"cell_type":"markdown","source":"# Testing model without fine-tuning","metadata":{}},{"cell_type":"code","source":"# # For clearing GPU memory\n# import gc\n\n# del [\n#     model, \n#     tokenizer, \n# ]\n# # del [df, X_train, X_eval]","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:53.137816Z","iopub.execute_input":"2024-10-17T04:18:53.138608Z","iopub.status.idle":"2024-10-17T04:18:53.142598Z","shell.execute_reply.started":"2024-10-17T04:18:53.138558Z","shell.execute_reply":"2024-10-17T04:18:53.141564Z"},"trusted":true},"execution_count":43,"outputs":[]},{"cell_type":"code","source":"# for _ in range(100):\n#     torch.cuda.empty_cache()\n#     gc.collect()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:53.611568Z","iopub.execute_input":"2024-10-17T04:18:53.612649Z","iopub.status.idle":"2024-10-17T04:18:53.616415Z","shell.execute_reply.started":"2024-10-17T04:18:53.612593Z","shell.execute_reply":"2024-10-17T04:18:53.615627Z"},"trusted":true},"execution_count":44,"outputs":[]},{"cell_type":"code","source":"model_name = \"/kaggle/input/llama-3.1/transformers/8b-instruct/2\"\n# model_name = \"/kaggle/input/llama-3.2/transformers/3b-instruct/1\"\n# model_name = \"/kaggle/input/llama-3.1/transformers/405b-instruct/1\"\n\ncompute_dtype = getattr(torch, \"float16\")\n\nbnb_config = BitsAndBytesConfig(\n    load_in_4bit=True, \n    bnb_4bit_quant_type=\"nf4\", \n    bnb_4bit_compute_dtype=compute_dtype,\n    bnb_4bit_use_double_quant=True,\n)\n\nmodel = AutoModelForCausalLM.from_pretrained(\n    model_name,\n    device_map=device,\n    torch_dtype=compute_dtype,\n    quantization_config=bnb_config, \n)\n\nmodel.config.use_cache = False\nmodel.config.pretraining_tp = 1\n\ntokenizer = AutoTokenizer.from_pretrained(model_name, \n                                          trust_remote_code=True,\n                                         )\ntokenizer.pad_token = tokenizer.eos_token\ntokenizer.padding_side = \"right\"\n\nmodel, tokenizer = setup_chat_format(model, tokenizer)","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:18:53.971361Z","iopub.execute_input":"2024-10-17T04:18:53.972203Z","iopub.status.idle":"2024-10-17T04:19:12.841639Z","shell.execute_reply.started":"2024-10-17T04:18:53.97216Z","shell.execute_reply":"2024-10-17T04:19:12.84079Z"},"trusted":true},"execution_count":45,"outputs":[{"output_type":"display_data","data":{"text/plain":"Loading checkpoint shards:   0%|          | 0/4 [00:00<?, ?it/s]","application/vnd.jupyter.widget-view+json":{"version_major":2,"version_minor":0,"model_id":"8d088b3986f144d9aaca39523f649429"}},"metadata":{}}]},{"cell_type":"code","source":"print(len(tokenizer))  # Check vocabulary size","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:19:12.843169Z","iopub.execute_input":"2024-10-17T04:19:12.843508Z","iopub.status.idle":"2024-10-17T04:19:12.871797Z","shell.execute_reply.started":"2024-10-17T04:19:12.843472Z","shell.execute_reply":"2024-10-17T04:19:12.870894Z"},"trusted":true},"execution_count":46,"outputs":[{"name":"stdout","text":"128258\n","output_type":"stream"}]},{"cell_type":"code","source":"def predict(df_data, model, tokenizer, debug=False):\n    y_pred = []\n    aspects = ['service', 'cleanliness', 'food', 'price', 'others']\n    \n    for i in tqdm(range(len(df_data))):\n        review = df_data[i]['review']\n        prompt = generate_prompt(review)\n        \n        pipe = pipeline(task=\"text-generation\", \n                        model=model, \n                        tokenizer=tokenizer, \n                        max_new_tokens=21, \n                        temperature=0.07,\n                       )\n        try:\n            result = pipe(prompt)\n            generated_text = result[0]['generated_text']\n            \n            # Extract only the part that matches our expected format\n            answer = re.search(r'Analysis:[\\s\\S]*', generated_text)\n            if answer:\n                answer = answer.group(0)\n            else:\n                raise ValueError(\"Unexpected output format\")\n            \n            aspect_sentiments = {}\n            \n            # Updated regex to capture scores more accurately, even with partial output\n            matches = re.findall(r'(\\w+)=\\[?(\\d)\\]?', answer)\n            \n            for match in matches:\n                aspect, score = match\n                aspect = aspect.lower()\n                score = int(score)\n                if aspect in aspects and score in [0, 1, 2, 3]:\n                    aspect_sentiments[aspect] = score\n            \n            # If no aspects were found or all aspects have zero scores, it's treated as a brief/generic review\n            if not aspect_sentiments or all(score == 0 for score in aspect_sentiments.values()):\n                aspect_sentiments = {aspect: 0 for aspect in aspects}\n            \n            # Fill in any missing aspects with 0\n            for aspect in aspects:\n                if aspect not in aspect_sentiments:\n                    aspect_sentiments[aspect] = 0\n            \n            y_pred.append(aspect_sentiments)\n            \n            if debug:\n                print(f\"\\nItem {i}:\")\n                print(f\"Review: {review}\")\n                print(f\"Generated: {answer}\")\n                print(f\"Parsed sentiments: {aspect_sentiments}\")\n        \n        except Exception as e:\n            print(f\"Error processing item {i}: {str(e)}\")\n            y_pred.append({aspect: 0 for aspect in aspects})\n    \n    return y_pred","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:20:18.658537Z","iopub.execute_input":"2024-10-17T04:20:18.659452Z","iopub.status.idle":"2024-10-17T04:20:18.671487Z","shell.execute_reply.started":"2024-10-17T04:20:18.659412Z","shell.execute_reply":"2024-10-17T04:20:18.670455Z"},"trusted":true},"execution_count":51,"outputs":[]},{"cell_type":"code","source":"y_pred = predict(df_data, model, tokenizer, debug=False)","metadata":{"execution":{"iopub.status.busy":"2024-10-17T04:20:34.402003Z","iopub.execute_input":"2024-10-17T04:20:34.40241Z","iopub.status.idle":"2024-10-17T05:57:01.171601Z","shell.execute_reply.started":"2024-10-17T04:20:34.40237Z","shell.execute_reply":"2024-10-17T05:57:01.170572Z"},"trusted":true},"execution_count":53,"outputs":[{"name":"stderr","text":"100%|██████████| 2374/2374 [1:36:26<00:00,  2.44s/it]\n","output_type":"stream"}]},{"cell_type":"code","source":"# Convert y_pred to a DataFrame\ny_pred_df = pd.DataFrame(y_pred)\n\n# Reset index of X_test to ensure it aligns with y_pred_df\nfiltered_df = filtered_df.reset_index(drop=True)\n\n# Concatenate X_test with y_pred_df\nresult_df = pd.concat([filtered_df, y_pred_df], axis=1)","metadata":{"execution":{"iopub.status.busy":"2024-10-17T06:06:15.063566Z","iopub.execute_input":"2024-10-17T06:06:15.064457Z","iopub.status.idle":"2024-10-17T06:06:15.080452Z","shell.execute_reply.started":"2024-10-17T06:06:15.064418Z","shell.execute_reply":"2024-10-17T06:06:15.079449Z"},"trusted":true},"execution_count":57,"outputs":[]},{"cell_type":"code","source":"# Display the first few rows of the result\nresult_df.head()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T06:06:16.233381Z","iopub.execute_input":"2024-10-17T06:06:16.23411Z","iopub.status.idle":"2024-10-17T06:06:16.251077Z","shell.execute_reply.started":"2024-10-17T06:06:16.23407Z","shell.execute_reply":"2024-10-17T06:06:16.250074Z"},"trusted":true},"execution_count":58,"outputs":[{"execution_count":58,"output_type":"execute_result","data":{"text/plain":"   reviewer_id  store_name              category  \\\n0           19  McDonald's  Fast food restaurant   \n1           20  McDonald's  Fast food restaurant   \n2           21  McDonald's  Fast food restaurant   \n3           24  McDonald's  Fast food restaurant   \n4           32  McDonald's  Fast food restaurant   \n\n                                       store_address  latitude   longitude  \\\n0  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n1  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n2  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n3  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n4  13749 US-183 Hwy, Austin, TX 78750, United States  30.460718 -97.792874   \n\n  rating_count    review_time  \\\n0        1,240   7 months ago   \n1        1,240   7 months ago   \n2        1,240  10 months ago   \n3        1,240   8 months ago   \n4        1,240   9 months ago   \n\n                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                    review  \\\n0                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           I am not happy at all today I went and bought 4 Strawberry Banana smoothies and and 2 hash browns and my smoothies tasted like straight ice and water smh And I am very disappointed in there service my smoothie should never taste like that never especially the fact I bought it for not only me but my kids I need someone to contact me from corporate or something I am not happy at all waste of money   \n1                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                            The staff seems to always be different which makes ordering sometimes a little difficult because they are in a state of perpetual learning which can also lead to waiting a bit longer but the lobby is always clean as well as the bathrooms   \n2  Cant believe experience with order taker Went from drive thru and employee directly asks what do you want He keeps rushing while taking order with no other cars placing order We wanted to add one more item and asked for a minute He keeps asking what else When I said we are ready  no response There were no other car placing order that time either I told again we are ready responded after a pause Again what do you want excuse me who talks to customers like this \\nI asked for his name to put a review and he flat out denied I asked for receipt before handing my credit card He swipes it and returns back my card I m waiting for a receipt with my family in the car He starts inputing another order on his tablet Waited for sometime and then I waive my hand several times He opens window and said you can get receipt from next window Excuse me \\nWent to next window  picked up food asked for Manager and got his name JAMAR My receipt states 08062022 720pm Order 05\\nI didnt want to complain but this was intolerable   \n3                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                              This is consistently the worst meal you will pay for Microwaved nuggets cold fries and a 30 minute wait What more can you ask for Dont go here ever Your better off going hungry or giving the meal you payed for to the homeless by the donation box behind the drive thru   \n4                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       This location is wack\\nThey will literally stop taking orders and make people sit in the parking lot for like 12 minutes until 300 am so that they dont get to order dinner food anymore\\nWe have gone multiple times right after leaving the pinballz arcade just to sit in the parking lot in the DriveThru line waiting while they do nothing but wait for the order menu to change over\\nLook McDonalds if I wanted a breakfast deluxe I wouldnt have shown up at 230 in the morning thank you   \n\n    rating sentiment                actual_date  service  cleanliness  food  \\\n0   1 star  negative 2024-03-21 04:07:20.534762        1            0     1   \n1  4 stars  positive 2024-03-21 04:07:20.534769        1            3     0   \n2   1 star  negative 2023-12-22 04:07:20.534777        1            0     0   \n3   1 star  negative 2024-02-20 04:07:20.534799        1            1     1   \n4  2 stars  negative 2024-01-21 04:07:20.534858        1            0     0   \n\n   price  others  \n0      0       1  \n1      0       0  \n2      0       1  \n3      1       1  \n4      0       1  ","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>reviewer_id</th>\n      <th>store_name</th>\n      <th>category</th>\n      <th>store_address</th>\n      <th>latitude</th>\n      <th>longitude</th>\n      <th>rating_count</th>\n      <th>review_time</th>\n      <th>review</th>\n      <th>rating</th>\n      <th>sentiment</th>\n      <th>actual_date</th>\n      <th>service</th>\n      <th>cleanliness</th>\n      <th>food</th>\n      <th>price</th>\n      <th>others</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>0</th>\n      <td>19</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>7 months ago</td>\n      <td>I am not happy at all today I went and bought 4 Strawberry Banana smoothies and and 2 hash browns and my smoothies tasted like straight ice and water smh And I am very disappointed in there service my smoothie should never taste like that never especially the fact I bought it for not only me but my kids I need someone to contact me from corporate or something I am not happy at all waste of money</td>\n      <td>1 star</td>\n      <td>negative</td>\n      <td>2024-03-21 04:07:20.534762</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>1</th>\n      <td>20</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>7 months ago</td>\n      <td>The staff seems to always be different which makes ordering sometimes a little difficult because they are in a state of perpetual learning which can also lead to waiting a bit longer but the lobby is always clean as well as the bathrooms</td>\n      <td>4 stars</td>\n      <td>positive</td>\n      <td>2024-03-21 04:07:20.534769</td>\n      <td>1</td>\n      <td>3</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n    </tr>\n    <tr>\n      <th>2</th>\n      <td>21</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>10 months ago</td>\n      <td>Cant believe experience with order taker Went from drive thru and employee directly asks what do you want He keeps rushing while taking order with no other cars placing order We wanted to add one more item and asked for a minute He keeps asking what else When I said we are ready  no response There were no other car placing order that time either I told again we are ready responded after a pause Again what do you want excuse me who talks to customers like this \\nI asked for his name to put a review and he flat out denied I asked for receipt before handing my credit card He swipes it and returns back my card I m waiting for a receipt with my family in the car He starts inputing another order on his tablet Waited for sometime and then I waive my hand several times He opens window and said you can get receipt from next window Excuse me \\nWent to next window  picked up food asked for Manager and got his name JAMAR My receipt states 08062022 720pm Order 05\\nI didnt want to complain but this was intolerable</td>\n      <td>1 star</td>\n      <td>negative</td>\n      <td>2023-12-22 04:07:20.534777</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>3</th>\n      <td>24</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>8 months ago</td>\n      <td>This is consistently the worst meal you will pay for Microwaved nuggets cold fries and a 30 minute wait What more can you ask for Dont go here ever Your better off going hungry or giving the meal you payed for to the homeless by the donation box behind the drive thru</td>\n      <td>1 star</td>\n      <td>negative</td>\n      <td>2024-02-20 04:07:20.534799</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n      <td>1</td>\n    </tr>\n    <tr>\n      <th>4</th>\n      <td>32</td>\n      <td>McDonald's</td>\n      <td>Fast food restaurant</td>\n      <td>13749 US-183 Hwy, Austin, TX 78750, United States</td>\n      <td>30.460718</td>\n      <td>-97.792874</td>\n      <td>1,240</td>\n      <td>9 months ago</td>\n      <td>This location is wack\\nThey will literally stop taking orders and make people sit in the parking lot for like 12 minutes until 300 am so that they dont get to order dinner food anymore\\nWe have gone multiple times right after leaving the pinballz arcade just to sit in the parking lot in the DriveThru line waiting while they do nothing but wait for the order menu to change over\\nLook McDonalds if I wanted a breakfast deluxe I wouldnt have shown up at 230 in the morning thank you</td>\n      <td>2 stars</td>\n      <td>negative</td>\n      <td>2024-01-21 04:07:20.534858</td>\n      <td>1</td>\n      <td>0</td>\n      <td>0</td>\n      <td>0</td>\n      <td>1</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"result_df.info()","metadata":{"execution":{"iopub.status.busy":"2024-10-17T06:06:37.410548Z","iopub.execute_input":"2024-10-17T06:06:37.410957Z","iopub.status.idle":"2024-10-17T06:06:37.425085Z","shell.execute_reply.started":"2024-10-17T06:06:37.41092Z","shell.execute_reply":"2024-10-17T06:06:37.423917Z"},"trusted":true},"execution_count":59,"outputs":[{"name":"stdout","text":"<class 'pandas.core.frame.DataFrame'>\nRangeIndex: 2374 entries, 0 to 2373\nData columns (total 17 columns):\n #   Column         Non-Null Count  Dtype         \n---  ------         --------------  -----         \n 0   reviewer_id    2374 non-null   int64         \n 1   store_name     2374 non-null   object        \n 2   category       2374 non-null   object        \n 3   store_address  2374 non-null   object        \n 4   latitude       2309 non-null   float64       \n 5   longitude      2309 non-null   float64       \n 6   rating_count   2374 non-null   object        \n 7   review_time    2374 non-null   object        \n 8   review         2374 non-null   object        \n 9   rating         2374 non-null   object        \n 10  sentiment      2374 non-null   object        \n 11  actual_date    2374 non-null   datetime64[ns]\n 12  service        2374 non-null   int64         \n 13  cleanliness    2374 non-null   int64         \n 14  food           2374 non-null   int64         \n 15  price          2374 non-null   int64         \n 16  others         2374 non-null   int64         \ndtypes: datetime64[ns](1), float64(2), int64(6), object(8)\nmemory usage: 315.4+ KB\n","output_type":"stream"}]},{"cell_type":"code","source":"# List of columns to analyze\ncolumns = ['service', 'cleanliness', 'food', 'price', 'others']\n\n# Pivoting data based on sentiment\nsummary_df = result_df.groupby('sentiment')[columns].apply(lambda x: x.apply(pd.Series.value_counts).reindex([0, 1, 2, 3], fill_value=0))\n\n# Reformat the result for better readability\nsummary_df = summary_df.stack(level=0).unstack().fillna(0)\n\n# Add a row subtotal for each sentiment\nsummary_df['Total'] = summary_df.sum(axis=1)\n\n# Add a column subtotal for each category (0, 1, 2, 3 types across sentiments)\nsummary_df.loc['Total'] = summary_df.sum(axis=0)\n\nsummary_df","metadata":{"execution":{"iopub.status.busy":"2024-10-17T06:18:22.285398Z","iopub.execute_input":"2024-10-17T06:18:22.286141Z","iopub.status.idle":"2024-10-17T06:18:22.344368Z","shell.execute_reply.started":"2024-10-17T06:18:22.286098Z","shell.execute_reply":"2024-10-17T06:18:22.343037Z"},"trusted":true},"execution_count":74,"outputs":[{"execution_count":74,"output_type":"execute_result","data":{"text/plain":"               service  cleanliness    food   price  others    Total\n(negative, 0)    206.0        882.0   619.0   944.0   151.0   2802.0\n(negative, 1)    766.0        105.0   357.0    47.0   829.0   2104.0\n(negative, 2)      0.0          4.0     3.0     0.0     1.0      8.0\n(negative, 3)     20.0          1.0    13.0     1.0    11.0     46.0\n(neutral, 0)     150.0        268.0   203.0   281.0   158.0   1060.0\n(neutral, 1)     128.0         27.0    75.0    19.0   130.0    379.0\n(neutral, 2)       2.0          2.0     7.0     1.0     4.0     16.0\n(neutral, 3)      25.0          8.0    20.0     4.0    13.0     70.0\n(positive, 0)    406.0        830.0   762.0  1033.0   563.0   3594.0\n(positive, 1)     98.0         12.0    49.0    13.0   183.0    355.0\n(positive, 2)      1.0          3.0    18.0     2.0    21.0     45.0\n(positive, 3)    572.0        232.0   248.0    29.0   310.0   1391.0\nTotal           2374.0       2374.0  2374.0  2374.0  2374.0  11870.0","text/html":"<div>\n<style scoped>\n    .dataframe tbody tr th:only-of-type {\n        vertical-align: middle;\n    }\n\n    .dataframe tbody tr th {\n        vertical-align: top;\n    }\n\n    .dataframe thead th {\n        text-align: right;\n    }\n</style>\n<table border=\"1\" class=\"dataframe\">\n  <thead>\n    <tr style=\"text-align: right;\">\n      <th></th>\n      <th>service</th>\n      <th>cleanliness</th>\n      <th>food</th>\n      <th>price</th>\n      <th>others</th>\n      <th>Total</th>\n    </tr>\n  </thead>\n  <tbody>\n    <tr>\n      <th>(negative, 0)</th>\n      <td>206.0</td>\n      <td>882.0</td>\n      <td>619.0</td>\n      <td>944.0</td>\n      <td>151.0</td>\n      <td>2802.0</td>\n    </tr>\n    <tr>\n      <th>(negative, 1)</th>\n      <td>766.0</td>\n      <td>105.0</td>\n      <td>357.0</td>\n      <td>47.0</td>\n      <td>829.0</td>\n      <td>2104.0</td>\n    </tr>\n    <tr>\n      <th>(negative, 2)</th>\n      <td>0.0</td>\n      <td>4.0</td>\n      <td>3.0</td>\n      <td>0.0</td>\n      <td>1.0</td>\n      <td>8.0</td>\n    </tr>\n    <tr>\n      <th>(negative, 3)</th>\n      <td>20.0</td>\n      <td>1.0</td>\n      <td>13.0</td>\n      <td>1.0</td>\n      <td>11.0</td>\n      <td>46.0</td>\n    </tr>\n    <tr>\n      <th>(neutral, 0)</th>\n      <td>150.0</td>\n      <td>268.0</td>\n      <td>203.0</td>\n      <td>281.0</td>\n      <td>158.0</td>\n      <td>1060.0</td>\n    </tr>\n    <tr>\n      <th>(neutral, 1)</th>\n      <td>128.0</td>\n      <td>27.0</td>\n      <td>75.0</td>\n      <td>19.0</td>\n      <td>130.0</td>\n      <td>379.0</td>\n    </tr>\n    <tr>\n      <th>(neutral, 2)</th>\n      <td>2.0</td>\n      <td>2.0</td>\n      <td>7.0</td>\n      <td>1.0</td>\n      <td>4.0</td>\n      <td>16.0</td>\n    </tr>\n    <tr>\n      <th>(neutral, 3)</th>\n      <td>25.0</td>\n      <td>8.0</td>\n      <td>20.0</td>\n      <td>4.0</td>\n      <td>13.0</td>\n      <td>70.0</td>\n    </tr>\n    <tr>\n      <th>(positive, 0)</th>\n      <td>406.0</td>\n      <td>830.0</td>\n      <td>762.0</td>\n      <td>1033.0</td>\n      <td>563.0</td>\n      <td>3594.0</td>\n    </tr>\n    <tr>\n      <th>(positive, 1)</th>\n      <td>98.0</td>\n      <td>12.0</td>\n      <td>49.0</td>\n      <td>13.0</td>\n      <td>183.0</td>\n      <td>355.0</td>\n    </tr>\n    <tr>\n      <th>(positive, 2)</th>\n      <td>1.0</td>\n      <td>3.0</td>\n      <td>18.0</td>\n      <td>2.0</td>\n      <td>21.0</td>\n      <td>45.0</td>\n    </tr>\n    <tr>\n      <th>(positive, 3)</th>\n      <td>572.0</td>\n      <td>232.0</td>\n      <td>248.0</td>\n      <td>29.0</td>\n      <td>310.0</td>\n      <td>1391.0</td>\n    </tr>\n    <tr>\n      <th>Total</th>\n      <td>2374.0</td>\n      <td>2374.0</td>\n      <td>2374.0</td>\n      <td>2374.0</td>\n      <td>2374.0</td>\n      <td>11870.0</td>\n    </tr>\n  </tbody>\n</table>\n</div>"},"metadata":{}}]},{"cell_type":"code","source":"result_df.to_csv('Labelled_7mth_11mth.csv')","metadata":{"execution":{"iopub.status.busy":"2024-10-17T06:21:37.223501Z","iopub.execute_input":"2024-10-17T06:21:37.223918Z","iopub.status.idle":"2024-10-17T06:21:37.282292Z","shell.execute_reply.started":"2024-10-17T06:21:37.223877Z","shell.execute_reply":"2024-10-17T06:21:37.281376Z"},"trusted":true},"execution_count":75,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}