{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "16fd83c7-9f91-40ab-ac15-57b02a63b7f4",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "os.environ['HF_HOME'] = \"/scratch/tar3kh/models/cache\"\n",
    "import torch \n",
    "\n",
    "from datasets import load_dataset #datasets is huggingface's dataset package\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import PIL"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f748fb12-da99-4702-bfb2-263e091fee14",
   "metadata": {},
   "source": [
    "## Synthetic Dataset (generating budgets)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "61502633-b04c-44fd-b39f-7803ef778205",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "size = 3000\n",
    "\n",
    "np.random.seed(60)\n",
    "Income_Randomizer = np.random.randint(29000,251000, size=size).astype(int)\n",
    "#print(Income_Randomizer)\n",
    "np.random.seed(60)\n",
    "\n",
    "Rent_Randomizer = np.random.randint(500,2500, size=size).astype(int)\n",
    "#print(Rent_Randomizer)\n",
    "np.random.seed(60)\n",
    "\n",
    "Car_Randomizer = np.random.randint(200,1000, size=size).astype(int)\n",
    "#print(Car_Randomizer)\n",
    "np.random.seed(60)\n",
    "\n",
    "Other_Randomizer = np.random.randint(200,600, size=size).astype(int)\n",
    "#print(Other_Randomizer)\n",
    "\n",
    "Example_promtps = []\n",
    "\n",
    "for x in range(len(Income_Randomizer)):\n",
    "    Example_promtps.append('I  have an income of about ' +\n",
    "                           str(Income_Randomizer[x]) +\n",
    "                           ' a year and my monthly expenses include ' +\n",
    "                           str(Rent_Randomizer[x]) +\n",
    "                           ' a month in rent and utilities, a ' +\n",
    "                           str(Car_Randomizer[x]) +\n",
    "                           ' car payment, $300 in food, and about ' +\n",
    "                           str(Other_Randomizer[x]) +\n",
    "                           ' a month in other expenses. Using python, can you create for me a budget spreadsheet and export it to excel?')\n",
    "\n",
    "#Example_promtps = ['I  have an income of about ' + str(Income_Randomizer[0]) + ' a year and my monthly expenses include ' + str(Rent_Randomizer[0]) + ' a month in rent and utilities, a ' + str(Car_Randomizer[0]) + ' car payment, $300 in food, and about ' + str(Other_Randomizer[0]) + ' a month in other expenses. Using python, can you create for me a budget spreadsheet and export it to excel?',]\n",
    "Example_outputs = []\n",
    "\n",
    "for x in range(len(Income_Randomizer)):\n",
    "    Example_outputs.append(''' import pandas as pd\n",
    "import openpyxl\n",
    "\n",
    "# Define income and expenses\n",
    "annual_income = '''+ str(Income_Randomizer[x])+'''\n",
    "monthly_income = annual_income / 12\n",
    "\n",
    "expenses = {\n",
    "    \"Rent & Utilities\": '''+ str(Rent_Randomizer[x] )+''',\n",
    "    \"Car Payment\": '''+ str(Car_Randomizer[x]) +''',\n",
    "    \"Food\": 300,\n",
    "    \"Other Expenses\": '''+ str(Other_Randomizer[x]) +'''\n",
    "}\n",
    "\n",
    "total_expenses = sum(expenses.values())\n",
    "net_savings = monthly_income - total_expenses\n",
    "\n",
    "# Create DataFrame\n",
    "budget_data = {\n",
    "    \"Category\": [\"Monthly Income\"] + list(expenses.keys()) + [\"Total Expenses\", \"Net Savings\"],\n",
    "    \"Amount ($)\": [monthly_income] + list(expenses.values()) + [total_expenses, net_savings]\n",
    "}\n",
    "\n",
    "df = pd.DataFrame(budget_data)\n",
    "\n",
    "# Save to Excel\n",
    "file_name = \"budget.xlsx\"\n",
    "df.to_excel(file_name, index=False, engine='openpyxl')\n",
    "\n",
    "print(f\"Budget spreadsheet saved as {file_name}\")''')\n",
    "\n",
    "df2 = pd.DataFrame({'question':Example_promtps,\n",
    "                    'response': Example_outputs})\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "79e98786-47d7-4a83-943a-7d5484bd4c2c",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df2['instruct'] = \"Q: \" + df2['question'] + \"\\n\\nA: \" + \"Lets think step by step.\" + df2['response']\n",
    "df2['question_1'] = \"Q: \" + df2['question'] + \"\\n\\nA: \" + \"Lets think step by step.\" "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "1199470a-790e-4c60-8bb5-15d7b64aa43a",
   "metadata": {
    "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>question</th>\n",
       "      <th>response</th>\n",
       "      <th>instruct</th>\n",
       "      <th>question_1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>I  have an income of about 162325 a year and m...</td>\n",
       "      <td>import pandas as pd\\nimport openpyxl\\n\\n# Def...</td>\n",
       "      <td>Q: I  have an income of about 162325 a year an...</td>\n",
       "      <td>Q: I  have an income of about 162325 a year an...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>I  have an income of about 35543 a year and my...</td>\n",
       "      <td>import pandas as pd\\nimport openpyxl\\n\\n# Def...</td>\n",
       "      <td>Q: I  have an income of about 35543 a year and...</td>\n",
       "      <td>Q: I  have an income of about 35543 a year and...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>I  have an income of about 203179 a year and m...</td>\n",
       "      <td>import pandas as pd\\nimport openpyxl\\n\\n# Def...</td>\n",
       "      <td>Q: I  have an income of about 203179 a year an...</td>\n",
       "      <td>Q: I  have an income of about 203179 a year an...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>I  have an income of about 197008 a year and m...</td>\n",
       "      <td>import pandas as pd\\nimport openpyxl\\n\\n# Def...</td>\n",
       "      <td>Q: I  have an income of about 197008 a year an...</td>\n",
       "      <td>Q: I  have an income of about 197008 a year an...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>I  have an income of about 223681 a year and m...</td>\n",
       "      <td>import pandas as pd\\nimport openpyxl\\n\\n# Def...</td>\n",
       "      <td>Q: I  have an income of about 223681 a year an...</td>\n",
       "      <td>Q: I  have an income of about 223681 a year an...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            question  \\\n",
       "0  I  have an income of about 162325 a year and m...   \n",
       "1  I  have an income of about 35543 a year and my...   \n",
       "2  I  have an income of about 203179 a year and m...   \n",
       "3  I  have an income of about 197008 a year and m...   \n",
       "4  I  have an income of about 223681 a year and m...   \n",
       "\n",
       "                                            response  \\\n",
       "0   import pandas as pd\\nimport openpyxl\\n\\n# Def...   \n",
       "1   import pandas as pd\\nimport openpyxl\\n\\n# Def...   \n",
       "2   import pandas as pd\\nimport openpyxl\\n\\n# Def...   \n",
       "3   import pandas as pd\\nimport openpyxl\\n\\n# Def...   \n",
       "4   import pandas as pd\\nimport openpyxl\\n\\n# Def...   \n",
       "\n",
       "                                            instruct  \\\n",
       "0  Q: I  have an income of about 162325 a year an...   \n",
       "1  Q: I  have an income of about 35543 a year and...   \n",
       "2  Q: I  have an income of about 203179 a year an...   \n",
       "3  Q: I  have an income of about 197008 a year an...   \n",
       "4  Q: I  have an income of about 223681 a year an...   \n",
       "\n",
       "                                          question_1  \n",
       "0  Q: I  have an income of about 162325 a year an...  \n",
       "1  Q: I  have an income of about 35543 a year and...  \n",
       "2  Q: I  have an income of about 203179 a year an...  \n",
       "3  Q: I  have an income of about 197008 a year an...  \n",
       "4  Q: I  have an income of about 223681 a year an...  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "1b66310d-41e4-4592-8516-b35b635baead",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df2.to_csv('budget_dataset.csv', index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2fac1974-19a0-4853-bbe5-9867b57819ce",
   "metadata": {},
   "source": [
    "## Synthetic Dataset (Financial Goals)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "e4d8b755-25ec-472f-a8e4-52d153ff2f46",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "# datset set up\n",
    "\n",
    "size = 3000\n",
    "\n",
    "\n",
    "np.random.seed(60)\n",
    "short_term_goals = np.random.randint(1000,5000, size=size).astype(int)\n",
    "\n",
    "np.random.seed(60)\n",
    "medium_term_goals = np.random.randint(5000,10000, size=size).astype(int)\n",
    "\n",
    "np.random.seed(60)\n",
    "long_term_goals = np.random.randint(75000,200000, size=size).astype(int)\n",
    "\n",
    "# print(short_term_goals)\n",
    "# print(medium_term_goals)\n",
    "# print(long_term_goals)\n",
    "\n",
    "prompts = []\n",
    "\n",
    "for x in range(len(short_term_goals)):\n",
    "    prompts.append('My short term goal is to save for a $' +\n",
    "                   str(short_term_goals[x]) +\n",
    "                   ' vacation in the next year, my medium term goal is to save for down payment for a new car, around ' +\n",
    "                   str(medium_term_goals[x]) +\n",
    "                   ' in the next 2 or 3 years, and my long term goal is to save for a down payment for a house around ' +\n",
    "                   str(long_term_goals[x]) +\n",
    "                   ' in the next ten years, can you help me integrate these goals into my budget as well as where I should store these savings?')\n",
    "\n",
    "outputs = []\n",
    "for x in range(len(short_term_goals)):\n",
    "    outputs.append(''' 1. Short-Term Goal: $'''+ str(short_term_goals[x]) +''' Vacation (1 Year)\n",
    "Timeline: 12 months\n",
    "Monthly Savings Needed: '''+ str(short_term_goals[x]) + ''' / 12 = '''+ str((short_term_goals[x]/12).round()) +'''\n",
    "\n",
    "Best Storage Option: High-yield savings account (HYSA)\n",
    "Easy access\n",
    "Earns some interest\n",
    "Safe from market fluctuations,\n",
    "\n",
    "2. Medium-Term Goal: $'''+ str(medium_term_goals[x]) +'''  Car Down Payment (2–3 Years)\n",
    "Timeline Options:\n",
    "2 years (24 months) → $''' + str((medium_term_goals[x]/24).round()) + '''/month\n",
    "3 years (36 months) → $''' + str((medium_term_goals[x]/36).round()) + '''/month\n",
    "Best Storage Option: HYSA or conservative investment\n",
    "If comfortable with some risk, a mix of HYSA + conservative investments (e.g., CDs, bond ETFs)\n",
    "If risk-averse, keep it in an HYSA,\n",
    "\n",
    "3. Long-Term Goal: $'''+ str(long_term_goals[x]) +'''   House Down Payment (10 Years)\n",
    "Timeline: 120 months\n",
    "Monthly Savings Needed: '''+ str(long_term_goals[x]) + ''' / 120 = '''+ str((long_term_goals[x]/120).round()) +''' \n",
    "\n",
    "Best Storage Option: Investment account\n",
    "Given the long time horizon, investing in a mix of index funds (S&P 500, total stock market) + bonds could provide higher returns.\n",
    "Consider Roth IRA (if eligible) or brokerage account to allow tax-efficient growth.\n",
    "\n",
    "Summary of Total Savings Targets:\n",
    "Total Monthly Savings goal = $''' +str(((short_term_goals[x]/12)+(medium_term_goals[x]/36)+(long_term_goals[x]/120)).round()) +''' - $''' +str(((short_term_goals[x]/12)+(medium_term_goals[x]/24)+(long_term_goals[x]/120)).round()) +'''/month'''\n",
    "                  )\n",
    "    \n",
    "df3 = pd.DataFrame({'question':prompts,\n",
    "                    'response':outputs})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "f7f484a6-5cc5-4a77-a474-1fc921095dc2",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df3['instruct'] = \"Q: \" + df3['question'] + \"\\n\\nA: \" + \"Lets think step by step.\" + df3['response']\n",
    "df3['question_1'] = \"Q: \" + df3['question'] + \"\\n\\nA: \" + \"Lets think step by step.\" "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "63d3c1cd-943b-41b3-ab4f-85f11510eeca",
   "metadata": {
    "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>question</th>\n",
       "      <th>response</th>\n",
       "      <th>instruct</th>\n",
       "      <th>question_1</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>My short term goal is to save for a $3253 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $3253 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $3253 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $3253 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>My short term goal is to save for a $4137 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $4137 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $4137 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $4137 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>My short term goal is to save for a $4654 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $4654 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $4654 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $4654 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>My short term goal is to save for a $2418 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $2418 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $2418 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $2418 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>My short term goal is to save for a $3447 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $3447 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $3447 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $3447 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>My short term goal is to save for a $3147 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $3147 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $3147 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $3147 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>My short term goal is to save for a $1072 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $1072 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $1072 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $1072 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>My short term goal is to save for a $3169 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $3169 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $3169 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $3169 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>My short term goal is to save for a $4985 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $4985 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $4985 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $4985 v...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>My short term goal is to save for a $3722 vaca...</td>\n",
       "      <td>1. Short-Term Goal: $3722 Vacation (1 Year)\\n...</td>\n",
       "      <td>Q: My short term goal is to save for a $3722 v...</td>\n",
       "      <td>Q: My short term goal is to save for a $3722 v...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                            question  \\\n",
       "0  My short term goal is to save for a $3253 vaca...   \n",
       "1  My short term goal is to save for a $4137 vaca...   \n",
       "2  My short term goal is to save for a $4654 vaca...   \n",
       "3  My short term goal is to save for a $2418 vaca...   \n",
       "4  My short term goal is to save for a $3447 vaca...   \n",
       "5  My short term goal is to save for a $3147 vaca...   \n",
       "6  My short term goal is to save for a $1072 vaca...   \n",
       "7  My short term goal is to save for a $3169 vaca...   \n",
       "8  My short term goal is to save for a $4985 vaca...   \n",
       "9  My short term goal is to save for a $3722 vaca...   \n",
       "\n",
       "                                            response  \\\n",
       "0   1. Short-Term Goal: $3253 Vacation (1 Year)\\n...   \n",
       "1   1. Short-Term Goal: $4137 Vacation (1 Year)\\n...   \n",
       "2   1. Short-Term Goal: $4654 Vacation (1 Year)\\n...   \n",
       "3   1. Short-Term Goal: $2418 Vacation (1 Year)\\n...   \n",
       "4   1. Short-Term Goal: $3447 Vacation (1 Year)\\n...   \n",
       "5   1. Short-Term Goal: $3147 Vacation (1 Year)\\n...   \n",
       "6   1. Short-Term Goal: $1072 Vacation (1 Year)\\n...   \n",
       "7   1. Short-Term Goal: $3169 Vacation (1 Year)\\n...   \n",
       "8   1. Short-Term Goal: $4985 Vacation (1 Year)\\n...   \n",
       "9   1. Short-Term Goal: $3722 Vacation (1 Year)\\n...   \n",
       "\n",
       "                                            instruct  \\\n",
       "0  Q: My short term goal is to save for a $3253 v...   \n",
       "1  Q: My short term goal is to save for a $4137 v...   \n",
       "2  Q: My short term goal is to save for a $4654 v...   \n",
       "3  Q: My short term goal is to save for a $2418 v...   \n",
       "4  Q: My short term goal is to save for a $3447 v...   \n",
       "5  Q: My short term goal is to save for a $3147 v...   \n",
       "6  Q: My short term goal is to save for a $1072 v...   \n",
       "7  Q: My short term goal is to save for a $3169 v...   \n",
       "8  Q: My short term goal is to save for a $4985 v...   \n",
       "9  Q: My short term goal is to save for a $3722 v...   \n",
       "\n",
       "                                          question_1  \n",
       "0  Q: My short term goal is to save for a $3253 v...  \n",
       "1  Q: My short term goal is to save for a $4137 v...  \n",
       "2  Q: My short term goal is to save for a $4654 v...  \n",
       "3  Q: My short term goal is to save for a $2418 v...  \n",
       "4  Q: My short term goal is to save for a $3447 v...  \n",
       "5  Q: My short term goal is to save for a $3147 v...  \n",
       "6  Q: My short term goal is to save for a $1072 v...  \n",
       "7  Q: My short term goal is to save for a $3169 v...  \n",
       "8  Q: My short term goal is to save for a $4985 v...  \n",
       "9  Q: My short term goal is to save for a $3722 v...  "
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df3.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "2a4deb48-724f-46a4-8e55-4e41f648af6e",
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "df3.to_csv('goals_dataset.csv', index=False)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "313cd80a-c7be-489c-9c70-30885c7e614a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "20de1e90-1c13-486a-b5db-09c957622a69",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "llm_course_2",
   "language": "python",
   "name": "llm_course_2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.11"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}