BP2024/preprocessing/chunking.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data chunking for effectiveness\n",
    "\n",
    "In our data, facebook user called Robert Fico has a lot of samples.\n",
    "For efficiency, this notebook chunks those data in 4 parts."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import json"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### JSONL file loading and creation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def load_jsonl(file_path):\n",
    "    with open(file_path, 'r', encoding='utf-8') as file:\n",
    "        return [json.loads(line) for line in file]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_jsonl(filename, new_dataset):\n",
    "    with open(f'{filename}l', 'w') as jsonl_file:\n",
    "        for item in new_dataset:\n",
    "            jsonl_file.write(json.dumps(item) + '\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "fico = load_jsonl('jsonl_data/robert_fico_data.jsonl')"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Split data into 4 parts equal parts"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "135155"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "num_samples = len(fico)\n",
    "chunk_size = int(num_samples / 4)\n",
    "\n",
    "num_samples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "chunk_size * 4 == num_samples  # we have lost one sample, because our dataset has odd number of samples"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Actual chunking algorithm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "chunk_arr = []\n",
    "for chunks in range(0, 4):\n",
    "    chunk_arr.append(\n",
    "        fico[chunk_size * chunks: chunk_size * (chunks + 1)]\n",
    "    )"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Write chunked data to disk in a for loop"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "for index, data in enumerate(chunk_arr):\n",
    "    create_jsonl(f'jsonl_data/fico_chunk_{index}.json', data)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "sentiment",
   "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": "3.9.18"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}