notebooks/growth_model_generate_data.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "7494ff17-adba-46cc-838b-43ef67ab0ddf",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T00:50:37.164970Z",
     "iopub.status.busy": "2024-03-11T00:50:37.164581Z",
     "iopub.status.idle": "2024-03-11T00:50:37.169658Z",
     "shell.execute_reply": "2024-03-11T00:50:37.168895Z",
     "shell.execute_reply.started": "2024-03-11T00:50:37.164946Z"
    }
   },
   "source": [
    "# Overview\n",
    "\n",
    "This Notebook demonstrates how to use OpenAI's GPT-3 language model to generate a dataset containing account names, dates, and revenue for forecasting time series growth.\n",
    "\n",
    "The dataset is generated based on a prompt provided to the GPT-3 model, and it includes a variety of revenue growth patterns over time.\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "123779e8-009c-43c9-8b09-4ea428d2e8bd",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T00:51:29.876967Z",
     "iopub.status.busy": "2024-03-11T00:51:29.876245Z",
     "iopub.status.idle": "2024-03-11T00:51:29.879857Z",
     "shell.execute_reply": "2024-03-11T00:51:29.879252Z",
     "shell.execute_reply.started": "2024-03-11T00:51:29.876939Z"
    }
   },
   "source": [
    "## Import Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "e0de21a0-ff8e-413c-b7bb-03a6ba279858",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:21:32.521516Z",
     "iopub.status.busy": "2024-03-11T18:21:32.521114Z",
     "iopub.status.idle": "2024-03-11T18:21:32.525158Z",
     "shell.execute_reply": "2024-03-11T18:21:32.524563Z",
     "shell.execute_reply.started": "2024-03-11T18:21:32.521489Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "from dotenv import load_dotenv\n",
    "import openai\n",
    "import pandas as pd\n",
    "from dateutil import parser as date_parser\n",
    "import re"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f3e816ea-a5b1-4e27-aa56-236499f4f7d9",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T00:52:33.861578Z",
     "iopub.status.busy": "2024-03-11T00:52:33.860863Z",
     "iopub.status.idle": "2024-03-11T00:52:33.864680Z",
     "shell.execute_reply": "2024-03-11T00:52:33.863857Z",
     "shell.execute_reply.started": "2024-03-11T00:52:33.861535Z"
    }
   },
   "source": [
    "## Setup openAI api key"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "441c3650-446a-4234-a591-dc3f00cc773b",
   "metadata": {},
   "source": [
    "Before running this code, make sure to set up your OpenAI API key in the environment file (.env). Follow the instructions here: https://platform.openai.com/docs/quickstart?context=python"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "4c470127-7cb5-4e43-9eab-b31700625ff5",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:21:33.716109Z",
     "iopub.status.busy": "2024-03-11T18:21:33.715723Z",
     "iopub.status.idle": "2024-03-11T18:21:33.720295Z",
     "shell.execute_reply": "2024-03-11T18:21:33.719735Z",
     "shell.execute_reply.started": "2024-03-11T18:21:33.716080Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "\n",
    "load_dotenv('.env')  \n",
    "openai_api_key = os.getenv('OPENAI_API_KEY') \n",
    "openai.api_key = openai_api_key\n"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ef69a466-d80e-4a79-86ae-754c1a836bd4",
   "metadata": {},
   "source": [
    "## Define Prompt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "ec1438d3-a506-4f22-ac5d-aa0091aba217",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:21:34.855004Z",
     "iopub.status.busy": "2024-03-11T18:21:34.854624Z",
     "iopub.status.idle": "2024-03-11T18:21:34.858497Z",
     "shell.execute_reply": "2024-03-11T18:21:34.857934Z",
     "shell.execute_reply.started": "2024-03-11T18:21:34.854977Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "prompt = \"\"\"\n",
    "Generate a dataset containing account names, dates, and revenue for forecasting time series growth.\n",
    "\n",
    "1. Define account name : [Account Name]\n",
    "\n",
    "2. Generate dates from January 1, 2023, to July 1, 2023, with weekly intervals for each account.\n",
    "\n",
    "3. Assign revenue values for each account and date:\n",
    "   - For each account:\n",
    "     - Randomly vary the revenue values for each date:\n",
    "       - For each date:\n",
    "         - The revenue may:\n",
    "           - Increase over time.\n",
    "           - Decrease over time.\n",
    "           - Remain constant over time.\n",
    "\n",
    "4. Ensure that the generated dataset reflects a variety of revenue growth patterns over time.\n",
    "\n",
    "Please generate the dataset in CSV format with the following columns:\n",
    "- Account Name\n",
    "- Date\n",
    "- Revenue\n",
    "\"\"\"\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "ecef194e-40d4-47bb-ae4c-5d156bb5d814",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:21:36.212882Z",
     "iopub.status.busy": "2024-03-11T18:21:36.212484Z",
     "iopub.status.idle": "2024-03-11T18:27:04.463865Z",
     "shell.execute_reply": "2024-03-11T18:27:04.463202Z",
     "shell.execute_reply.started": "2024-03-11T18:21:36.212853Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "data=[]\n",
    "# Looping through each account\n",
    "for account_number in range(1, 51): # creating 50 accounts\n",
    "    account_name = f\"Account Name {account_number}\"\n",
    "    account_prompt = prompt.replace(\"[Account Name]\", account_name)\n",
    "    \n",
    "    # Request\n",
    "    response = openai.Completion.create(\n",
    "        engine=\"gpt-3.5-turbo-instruct\",  \n",
    "        prompt=account_prompt,\n",
    "        max_tokens=1000\n",
    "    )\n",
    "\n",
    "    # generate response\n",
    "    generated_text = response.choices[0].text.strip()\n",
    "\n",
    "    # Pre-process generated text, \n",
    "    for line in generated_text.split('\\n')[2:]:\n",
    "        try:\n",
    "            elements = line.strip().split(',')\n",
    "            date_str = elements[1].strip()\n",
    "            revenue = elements[2].strip()\n",
    "\n",
    "            # Parse the date string using dateutil.parser\n",
    "            date_obj = date_parser.parse(date_str)\n",
    "\n",
    "            # Check if the date falls within the specified range\n",
    "            if date_obj.year < 2023 or date_obj.year > 2023:\n",
    "                # Skip this entry if the date is outside the specified range\n",
    "                continue\n",
    "\n",
    "            # Convert the parsed date object back to a string in a uniform format\n",
    "            date_str_uniform = date_obj.strftime('%m/%d/%Y')\n",
    "\n",
    "            # Revenue format\n",
    "            if not re.match(r'^\\$\\d+(?:,\\d+)*(\\.\\d+)?$', revenue):\n",
    "                # Convert non-standard formats to integers\n",
    "                revenue = int(revenue)\n",
    "            else:\n",
    "                # Remove non-numeric characters and convert to integer\n",
    "                revenue = int(re.sub(r'[^\\d]', '', revenue))\n",
    "\n",
    "            # Append the data to the list\n",
    "            data.append({\n",
    "                'Account Name': account_name,\n",
    "                'Date': date_str_uniform,\n",
    "                'Revenue': revenue\n",
    "            })\n",
    "            \n",
    "        except Exception as e:\n",
    "            pass"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9cea9afd-11f7-400b-92fe-8f31dc3c351b",
   "metadata": {},
   "source": [
    "## Create dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "3e490257-9ba9-4698-938f-5fc52e0c4dbf",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:27:18.917284Z",
     "iopub.status.busy": "2024-03-11T18:27:18.916889Z",
     "iopub.status.idle": "2024-03-11T18:27:18.922274Z",
     "shell.execute_reply": "2024-03-11T18:27:18.921678Z",
     "shell.execute_reply.started": "2024-03-11T18:27:18.917257Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df = pd.DataFrame(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "f1333750-91ee-4a7a-8cfc-97a7f1a9807f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:27:20.320919Z",
     "iopub.status.busy": "2024-03-11T18:27:20.320183Z",
     "iopub.status.idle": "2024-03-11T18:27:20.328007Z",
     "shell.execute_reply": "2024-03-11T18:27:20.327363Z",
     "shell.execute_reply.started": "2024-03-11T18:27:20.320888Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "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>Account Name</th>\n",
       "      <th>Date</th>\n",
       "      <th>Revenue</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Account Name 1</td>\n",
       "      <td>01/01/2023</td>\n",
       "      <td>5000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Account Name 1</td>\n",
       "      <td>01/08/2023</td>\n",
       "      <td>5100</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Account Name 1</td>\n",
       "      <td>01/15/2023</td>\n",
       "      <td>5200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Account Name 1</td>\n",
       "      <td>01/22/2023</td>\n",
       "      <td>5400</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Account Name 1</td>\n",
       "      <td>01/29/2023</td>\n",
       "      <td>5400</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Account Name        Date  Revenue\n",
       "0  Account Name 1  01/01/2023     5000\n",
       "1  Account Name 1  01/08/2023     5100\n",
       "2  Account Name 1  01/15/2023     5200\n",
       "3  Account Name 1  01/22/2023     5400\n",
       "4  Account Name 1  01/29/2023     5400"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "1d22cb29-a845-4a84-848f-d8ce6d4fdf84",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:27:22.518189Z",
     "iopub.status.busy": "2024-03-11T18:27:22.517816Z",
     "iopub.status.idle": "2024-03-11T18:27:22.526284Z",
     "shell.execute_reply": "2024-03-11T18:27:22.525725Z",
     "shell.execute_reply.started": "2024-03-11T18:27:22.518165Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "<class 'pandas.core.frame.DataFrame'>\n",
      "RangeIndex: 1183 entries, 0 to 1182\n",
      "Data columns (total 3 columns):\n",
      " #   Column        Non-Null Count  Dtype \n",
      "---  ------        --------------  ----- \n",
      " 0   Account Name  1183 non-null   object\n",
      " 1   Date          1183 non-null   object\n",
      " 2   Revenue       1183 non-null   int64 \n",
      "dtypes: int64(1), object(2)\n",
      "memory usage: 27.9+ KB\n"
     ]
    }
   ],
   "source": [
    "df.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "a102ab25-36d2-4c97-8778-5090a77de878",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:27:23.105717Z",
     "iopub.status.busy": "2024-03-11T18:27:23.105291Z",
     "iopub.status.idle": "2024-03-11T18:27:23.109900Z",
     "shell.execute_reply": "2024-03-11T18:27:23.109171Z",
     "shell.execute_reply.started": "2024-03-11T18:27:23.105686Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df.rename(columns={'Account Name':'account_name', 'Date':'date', 'Revenue':'value'}, inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "bee1eb2b-ae1f-4a42-9e0e-fc46bf1b6ab3",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:27:23.958302Z",
     "iopub.status.busy": "2024-03-11T18:27:23.957654Z",
     "iopub.status.idle": "2024-03-11T18:27:23.965221Z",
     "shell.execute_reply": "2024-03-11T18:27:23.964639Z",
     "shell.execute_reply.started": "2024-03-11T18:27:23.958276Z"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "df['date']=pd.to_datetime(df['date'])\n",
    "df = df[(df['date'] >= '2023-01-01') & (df['date'] < '2023-08-01')] # making sure the date range is uniform"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "308575d2-2e84-4f70-8b24-c449c941ee0f",
   "metadata": {
    "execution": {
     "iopub.execute_input": "2024-03-11T18:27:28.085877Z",
     "iopub.status.busy": "2024-03-11T18:27:28.085439Z",
     "iopub.status.idle": "2024-03-11T18:27:28.095051Z",
     "shell.execute_reply": "2024-03-11T18:27:28.094404Z",
     "shell.execute_reply.started": "2024-03-11T18:27:28.085847Z"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CSV file generated successfully.\n"
     ]
    }
   ],
   "source": [
    "df.to_csv('../data/raw/generated_data.csv', index=False)\n",
    "\n",
    "print(\"CSV file generated successfully.\")"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.14"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}