{ "cells": [ { "cell_type": "markdown", "id": "84635d55-68a2-468b-ac09-9029ebdab55f", "metadata": { "id": "84635d55-68a2-468b-ac09-9029ebdab55f" }, "source": [ "# Accelerating large string data processing with cudf pandas accelerator mode (cudf.pandas)\n", "cuDF is a Python GPU DataFrame library (built on the Apache Arrow columnar memory format) for loading, joining, aggregating, filtering, and otherwise manipulating tabular data using a DataFrame style API in the style of pandas.\n", "\n", "cuDF now provides a pandas accelerator mode (`cudf.pandas`), allowing you to bring accelerated computing to your pandas workflows without requiring any code change.\n", "\n", "This notebook demonstrates how cuDF pandas accelerator mode can help accelerate processing of datasets with large string fields (4 GB+) processing by simply adding a `%load_ext` command. We have introduced this feature as part of our Rapids 24.08 release.\n", "\n", "**Author:** Allison Ding, Mitesh Patel
\n", "**Date:** October 3, 2025" ] }, { "cell_type": "markdown", "id": "bb8fe7ab-c055-40e9-897d-c62c72f28a16", "metadata": { "id": "bb8fe7ab-c055-40e9-897d-c62c72f28a16" }, "source": [ "# ⚠️ Verify your setup\n", "\n", "First, we'll verify that you are running with an NVIDIA GPU." ] }, { "cell_type": "code", "execution_count": 7, "id": "a88b8586-cfdd-4d31-9b4d-9be8508f7ba0", "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "a88b8586-cfdd-4d31-9b4d-9be8508f7ba0", "outputId": "18525b64-b34b-40e3-ed3a-1ad56ae794b5" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Fri Oct 3 23:16:52 2025 \n", "+-----------------------------------------------------------------------------------------+\n", "| NVIDIA-SMI 580.82.09 Driver Version: 580.82.09 CUDA Version: 13.0 |\n", "+-----------------------------------------+------------------------+----------------------+\n", "| GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC |\n", "| Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. |\n", "| | | MIG M. |\n", "|=========================================+========================+======================|\n", "| 0 NVIDIA GB10 Off | 0000000F:01:00.0 Off | N/A |\n", "| N/A 44C P0 10W / N/A | Not Supported | 0% Default |\n", "| | | N/A |\n", "+-----------------------------------------+------------------------+----------------------+\n", "\n", "+-----------------------------------------------------------------------------------------+\n", "| Processes: |\n", "| GPU GI CI PID Type Process name GPU Memory |\n", "| ID ID Usage |\n", "|=========================================================================================|\n", "| 0 N/A N/A 3405 G /usr/lib/xorg/Xorg 242MiB |\n", "| 0 N/A N/A 3562 G /usr/bin/gnome-shell 53MiB |\n", "| 0 N/A N/A 214921 C .../envs/rapids-25.10/bin/python 196MiB |\n", "+-----------------------------------------------------------------------------------------+\n" ] } ], "source": [ "!nvidia-smi # this should display information about available GPUs" ] }, { "cell_type": "markdown", "id": "5cd58071-4371-428b-8a02-9cd66e6cb91f", "metadata": { "id": "5cd58071-4371-428b-8a02-9cd66e6cb91f" }, "source": [ "# Download the data" ] }, { "cell_type": "markdown", "id": "9eb67713-7cf4-415a-bce7-ff4695862faa", "metadata": { "id": "9eb67713-7cf4-415a-bce7-ff4695862faa" }, "source": [ "## Overview\n", "The data we'll be working with summarizes job postings data that a developer working at a job listing firm might analyze to understand posting trends.\n", "\n", "We'll need to download a curated copy of this [Kaggle dataset](https://www.kaggle.com/datasets/asaniczka/1-3m-linkedin-jobs-and-skills-2024/data?select=job_summary.csv) directly from the kaggle API. \n", "\n", "**Data License and Terms**
\n", "As this dataset originates from a Kaggle dataset, it's governed by that dataset's license and terms of use, which is the Open Data Commons license. Review here:https://opendatacommons.org/licenses/by/1-0/index.html. For each dataset an user elects to use, the user is responsible for checking if the dataset license is fit for the intended purpose.\n", "\n", "**Are there restrictions on how I can use this data?
**\n", "For each dataset an user elects to use, the user is responsible for checking if the dataset license is fit for the intended purpose.\n", "\n", "## Get the Data\n", "First, [please follow these instructions from Kaggle to download and/or updating your Kaggle API token to get acces the dataset](https://www.kaggle.com/discussions/general/74235). \n", "\n", "Once generated, make sure to have the **kaggle.json** file in the same folder as the notebook\n", "\n", "Next, run this code below, which should also take 1-2 minutes:" ] }, { "cell_type": "code", "execution_count": 8, "id": "406838c6-267c-423e-82ab-ea13d5fa9c90", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Requirement already satisfied: kaggle in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (1.7.4.5)\n", "Requirement already satisfied: bleach in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (6.2.0)\n", "Requirement already satisfied: certifi>=14.05.14 in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (2025.8.3)\n", "Requirement already satisfied: charset-normalizer in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (3.4.3)\n", "Requirement already satisfied: idna in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (3.10)\n", "Requirement already satisfied: protobuf in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (6.32.1)\n", "Requirement already satisfied: python-dateutil>=2.5.3 in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (2.9.0.post0)\n", "Requirement already satisfied: python-slugify in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (8.0.4)\n", "Requirement already satisfied: requests in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (2.32.5)\n", "Requirement already satisfied: setuptools>=21.0.0 in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (80.9.0)\n", "Requirement already satisfied: six>=1.10 in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (1.17.0)\n", "Requirement already satisfied: text-unidecode in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (1.3)\n", "Requirement already satisfied: tqdm in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (4.67.1)\n", "Requirement already satisfied: urllib3>=1.15.1 in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (2.5.0)\n", "Requirement already satisfied: webencodings in /home/nvidia/miniconda3/envs/rapids-25.10/lib/python3.12/site-packages (from kaggle) (0.5.1)\n" ] } ], "source": [ "!pip install kaggle\n", "!mkdir -p ~/.kaggle\n", "!cp kaggle.json ~/.kaggle/\n", "!chmod 600 ~/.kaggle/kaggle.json" ] }, { "cell_type": "code", "execution_count": 18, "id": "3efacb3c-5f3d-4ff0-b32a-76bbb80b5f74", "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "3efacb3c-5f3d-4ff0-b32a-76bbb80b5f74", "outputId": "5fe4a878-cf57-44f9-e40e-ed413035b150" }, "outputs": [], "source": [ "# Download the dataset through kaggle API-\n", "!kaggle datasets download -d asaniczka/1-3m-linkedin-jobs-and-skills-2024\n", "#unzip the file to access contents\n", "!unzip 1-3m-linkedin-jobs-and-skills-2024.zip" ] }, { "cell_type": "markdown", "id": "2__ZMVe6LaBJ", "metadata": { "id": "2__ZMVe6LaBJ" }, "source": [ "# Analysis with cuDF Pandas" ] }, { "cell_type": "markdown", "id": "df47f304-2b30-4380-afd5-0613b63d103d", "metadata": {}, "source": [ "The magic command `%load_ext cudf.pandas` enables GPU acceleration for pandas data processing in a Jupyter notebook, allowing most pandas operations to automatically execute on NVIDIA GPUs for improved performance. \n", "\n", "With this extension loaded before importing pandas, your code can use standard pandas syntax while gaining the benefits of GPU speedup, automatically falling back to CPU execution for operations not supported on the GPU. This provides a seamless way to accelerate existing pandas workflows with zero code changes, especially for large data analytics tasks or machine learning preprocessing." ] }, { "cell_type": "code", "execution_count": 1, "id": "e5cd2520-30a6-41c1-b7c5-5abe0eb90d82", "metadata": {}, "outputs": [], "source": [ "%load_ext cudf.pandas" ] }, { "cell_type": "code", "execution_count": 2, "id": "eadb8d77-cb45-4c7c-ae9f-77e47a4f29b3", "metadata": { "id": "eadb8d77-cb45-4c7c-ae9f-77e47a4f29b3" }, "outputs": [], "source": [ "import pandas as pd\n", "import numpy as np" ] }, { "cell_type": "markdown", "id": "196268f2-6169-4ed7-a9e6-db9078caa6ab", "metadata": { "id": "196268f2-6169-4ed7-a9e6-db9078caa6ab" }, "source": [ "We'll run a piece of code to get a feel what GPU-acceleration brings to pandas workflows." ] }, { "cell_type": "code", "execution_count": null, "id": "ae3b6a16-ff72-4421-b43c-06c33f57ec12", "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "ae3b6a16-ff72-4421-b43c-06c33f57ec12", "outputId": "656acbf7-078f-42b3-832d-ad4e84e01c70" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 185 ms, sys: 2.08 s, total: 2.27 s\n", "Wall time: 2.95 s\n", "Dataset Size (in GB): 4.76\n" ] } ], "source": [ "%%time \n", "job_summary_df = pd.read_csv(\"job_summary.csv\", dtype=('str'))\n", "print(\"Dataset Size (in GB):\",round(job_summary_df.memory_usage(\n", " deep=True).sum()/(1024**3),2))" ] }, { "cell_type": "markdown", "id": "01c506e1-f135-4afb-8fc7-23e72c05d73c", "metadata": { "id": "01c506e1-f135-4afb-8fc7-23e72c05d73c" }, "source": [ "The same dataset takes about around 1.5 minutes to load with pandas. That's around **5x speedup** with no changes to the code!" ] }, { "cell_type": "markdown", "id": "d9d0a0e1-1d74-494d-bd12-b829f11eeede", "metadata": { "id": "d9d0a0e1-1d74-494d-bd12-b829f11eeede" }, "source": [ "Let's load the remaining two datasets as well:" ] }, { "cell_type": "code", "execution_count": 4, "id": "12e4cf7e-8824-4822-9d30-46b81ba2acd7", "metadata": { "colab": { "base_uri": "https://localhost:8080/" }, "id": "12e4cf7e-8824-4822-9d30-46b81ba2acd7", "outputId": "5ca1be17-09e3-40ab-928b-82176bf597bf" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 45.3 ms, sys: 199 ms, total: 244 ms\n", "Wall time: 354 ms\n" ] } ], "source": [ "%%time\n", "job_skills_df = pd.read_csv(\"job_skills.csv\", dtype=('str'))\n", "job_postings_df = pd.read_csv(\"linkedin_job_postings.csv\", dtype=('str'))" ] }, { "cell_type": "code", "execution_count": 38, "id": "13c8f9da-121f-4311-8a79-274425363e5e", "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 276 }, "id": "13c8f9da-121f-4311-8a79-274425363e5e", "outputId": "a73599c1-05b2-4f56-a190-c69c017bb330" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 4.46 ms, sys: 3.1 ms, total: 7.56 ms\n", "Wall time: 46.3 ms\n" ] }, { "data": { "text/plain": [ "0 957\n", "1 3816\n", "2 5314\n", "3 2774\n", "4 2749\n", "Name: summary_length, dtype: int32" ] }, "execution_count": 38, "metadata": {}, "output_type": "execute_result" } ], "source": [ "%%time\n", "job_summary_df['summary_length'] = job_summary_df['job_summary'].str.len()\n", "job_summary_df['summary_length'].head()" ] }, { "cell_type": "markdown", "id": "67b68792-5c64-4ebd-9d80-cf6ff55baeef", "metadata": { "id": "67b68792-5c64-4ebd-9d80-cf6ff55baeef" }, "source": [ "That was lightning fast! We went from around 10+ (with pandas) to a few milliseconds." ] }, { "cell_type": "code", "execution_count": 39, "id": "31e1cc84-debb-4da7-bc20-5c7139f786f7", "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 504 }, "id": "31e1cc84-debb-4da7-bc20-5c7139f786f7", "outputId": "2d89fc49-7e5b-41db-c25b-441d54480711" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 39.8 ms, sys: 30 ms, total: 69.8 ms\n", "Wall time: 211 ms\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
job_linklast_processed_timegot_summarygot_neris_being_workedjob_titlecompanyjob_locationfirst_seensearch_citysearch_countrysearch_positionjob_leveljob_typejob_summarysummary_length
0https://www.linkedin.com/jobs/view/account-exe...2024-01-21 07:12:29.00256+00ttfAccount Executive - Dispensing (NorCal/Norther...BDSan Diego, CA2024-01-15CoronadoUnited StatesColor MakerMid seniorOnsiteResponsibilities\\nJob Description Summary\\nJob...4602
1https://www.linkedin.com/jobs/view/registered-...2024-01-21 07:39:58.88137+00ttfRegistered Nurse - RN Care ManagerTrinity Health MINorton Shores, MI2024-01-14Grand HavenUnited StatesDirector Nursing ServiceMid seniorOnsiteEmployment Type:\\nFull time\\nShift:\\nDescripti...2950
2https://www.linkedin.com/jobs/view/restaurant-...2024-01-21 07:40:00.251126+00ttfRESTAURANT SUPERVISOR - THE FORKLIFTWasatch Adaptive SportsSandy, UT2024-01-14TooeleUnited StatesStand-InMid seniorOnsiteJob Details\\nDescription\\nWhat You'll Do\\nAs a...4571
3https://www.linkedin.com/jobs/view/independent...2024-01-21 07:40:00.308133+00ttfIndependent Real Estate AgentHoward Hanna | Rand RealtyEnglewood Cliffs, NJ2024-01-16PinehurstUnited StatesReal-Estate ClerkMid seniorOnsiteWho We Are\\nRand Realty is a family-owned brok...3944
4https://www.linkedin.com/jobs/view/group-unit-...2024-01-19 09:45:09.215838+00fffGroup/Unit Supervisor (Systems Support Manager...IRS, Office of Chief CounselChamblee, GA2024-01-17GadsdenUnited StatesSupervisor Travel-Information CenterMid seniorOnsiteNone<NA>
\n", "
" ], "text/plain": [ " job_link \\\n", "0 https://www.linkedin.com/jobs/view/account-exe... \n", "1 https://www.linkedin.com/jobs/view/registered-... \n", "2 https://www.linkedin.com/jobs/view/restaurant-... \n", "3 https://www.linkedin.com/jobs/view/independent... \n", "4 https://www.linkedin.com/jobs/view/group-unit-... \n", "\n", " last_processed_time got_summary got_ner is_being_worked \\\n", "0 2024-01-21 07:12:29.00256+00 t t f \n", "1 2024-01-21 07:39:58.88137+00 t t f \n", "2 2024-01-21 07:40:00.251126+00 t t f \n", "3 2024-01-21 07:40:00.308133+00 t t f \n", "4 2024-01-19 09:45:09.215838+00 f f f \n", "\n", " job_title \\\n", "0 Account Executive - Dispensing (NorCal/Norther... \n", "1 Registered Nurse - RN Care Manager \n", "2 RESTAURANT SUPERVISOR - THE FORKLIFT \n", "3 Independent Real Estate Agent \n", "4 Group/Unit Supervisor (Systems Support Manager... \n", "\n", " company job_location first_seen \\\n", "0 BD San Diego, CA 2024-01-15 \n", "1 Trinity Health MI Norton Shores, MI 2024-01-14 \n", "2 Wasatch Adaptive Sports Sandy, UT 2024-01-14 \n", "3 Howard Hanna | Rand Realty Englewood Cliffs, NJ 2024-01-16 \n", "4 IRS, Office of Chief Counsel Chamblee, GA 2024-01-17 \n", "\n", " search_city search_country search_position \\\n", "0 Coronado United States Color Maker \n", "1 Grand Haven United States Director Nursing Service \n", "2 Tooele United States Stand-In \n", "3 Pinehurst United States Real-Estate Clerk \n", "4 Gadsden United States Supervisor Travel-Information Center \n", "\n", " job_level job_type job_summary \\\n", "0 Mid senior Onsite Responsibilities\\nJob Description Summary\\nJob... \n", "1 Mid senior Onsite Employment Type:\\nFull time\\nShift:\\nDescripti... \n", "2 Mid senior Onsite Job Details\\nDescription\\nWhat You'll Do\\nAs a... \n", "3 Mid senior Onsite Who We Are\\nRand Realty is a family-owned brok... \n", "4 Mid senior Onsite None \n", "\n", " summary_length \n", "0 4602 \n", "1 2950 \n", "2 4571 \n", "3 3944 \n", "4 " ] }, "execution_count": 39, "metadata": {}, "output_type": "execute_result" } ], "source": [ "%%time\n", "df_merged=pd.merge(job_postings_df, job_summary_df, how=\"left\", on=\"job_link\")\n", "df_merged.head()" ] }, { "cell_type": "code", "execution_count": 40, "id": "0160a559-2b17-40a6-ad9d-34ce746236d0", "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 490 }, "id": "0160a559-2b17-40a6-ad9d-34ce746236d0", "outputId": "e397c28b-a90d-42d2-8a9a-4c6260c45b38" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 33.2 ms, sys: 17.3 ms, total: 50.6 ms\n", "Wall time: 120 ms\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
summary_length
companyjob_title
ClickJobs.ioAdolescent Behavioral Health Therapist - Substance Use Specialty (Entry Senior Level) Psychiatry23748.0
Mt. San Antonio CollegeChief, Police and Campus Safety22998.0
CareerBeaconAirside/Groundside Project Manager [Halifax International Airport Authority]22938.0
Tacoma Community CollegeAnthropology Professor - Part-time22790.0
IRS, Office of Chief CounselProgram Analyst (12-Month Roster)22774.0
.........
鴻海精密工業股份有限公司HR Specialist - Payroll & Benefit0.0
Material Planner0.0
RFQ Specialist0.0
Supply Chain Program Manager0.0
🌟Daniel-Scott Recruitment Ltd🌟IT Manager0.0
\n", "

801276 rows × 1 columns

\n", "
" ], "text/plain": [ " summary_length\n", "company job_title \n", "ClickJobs.io Adolescent Behavioral Health Therapist - Substa... 23748.0\n", "Mt. San Antonio College Chief, Police and Campus Safety 22998.0\n", "CareerBeacon Airside/Groundside Project Manager [Halifax Int... 22938.0\n", "Tacoma Community College Anthropology Professor - Part-time 22790.0\n", "IRS, Office of Chief Counsel Program Analyst (12-Month Roster) 22774.0\n", "... ...\n", "鴻海精密工業股份有限公司 HR Specialist - Payroll & Benefit 0.0\n", " Material Planner 0.0\n", " RFQ Specialist 0.0\n", " Supply Chain Program Manager 0.0\n", "🌟Daniel-Scott Recruitment Ltd🌟 IT Manager 0.0\n", "\n", "[801276 rows x 1 columns]" ] }, "execution_count": 40, "metadata": {}, "output_type": "execute_result" } ], "source": [ "%%time\n", "df_merged.groupby(['company',\"job_title\"]).agg({\n", " \"summary_length\":\"mean\"}).sort_values(by='summary_length', ascending = False).fillna(0)" ] }, { "cell_type": "markdown", "id": "IME4urGYQ3qS", "metadata": { "id": "IME4urGYQ3qS" }, "source": [ "We went down from around 5 seconds to less than a second here. This is in line with our speedups on other operations!" ] }, { "cell_type": "code", "execution_count": 41, "id": "adc00726-f151-41f4-8731-a1ce1f83eea2", "metadata": { "colab": { "base_uri": "https://localhost:8080/", "height": 458 }, "id": "adc00726-f151-41f4-8731-a1ce1f83eea2", "outputId": "46423696-b167-4ffe-bb3b-9de7f3e6d668" }, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "CPU times: user 13.7 ms, sys: 20.3 ms, total: 34 ms\n", "Wall time: 156 ms\n" ] }, { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
job_titlejob_locationsummary_length
0🔥Nurse Manager, Patient Services - Operating RoomLake George, NY7342.0
1🔥Behavioral Health RN 3 12sGlens Falls, NY2787.0
2🔥 Surgical Technologist - EveningsLake George, NY2920.0
3🔥 Physician Practice Clinical Lead RNSaratoga Springs, NY2945.0
4🔥 Physican Practice LPN - GreenLake George, NY2969.0
............
1104106\"Attorney\" (Gov Appt/Non-Merit) JobsKentucky, United States2427.0
1104107\"Accountant\"Shavano Park, TX1497.0
1104108\"Accountant\"Basking Ridge, NJ1073.0
1104109\"Accountant\"Austin, TX1993.0
1104110\"A\" Softball Coach - Central Middle SchoolEast Corinth, ME718.0
\n", "

1104111 rows × 3 columns

\n", "
" ], "text/plain": [ " job_title \\\n", "0 🔥Nurse Manager, Patient Services - Operating Room \n", "1 🔥Behavioral Health RN 3 12s \n", "2 🔥 Surgical Technologist - Evenings \n", "3 🔥 Physician Practice Clinical Lead RN \n", "4 🔥 Physican Practice LPN - Green \n", "... ... \n", "1104106 \"Attorney\" (Gov Appt/Non-Merit) Jobs \n", "1104107 \"Accountant\" \n", "1104108 \"Accountant\" \n", "1104109 \"Accountant\" \n", "1104110 \"A\" Softball Coach - Central Middle School \n", "\n", " job_location summary_length \n", "0 Lake George, NY 7342.0 \n", "1 Glens Falls, NY 2787.0 \n", "2 Lake George, NY 2920.0 \n", "3 Saratoga Springs, NY 2945.0 \n", "4 Lake George, NY 2969.0 \n", "... ... ... \n", "1104106 Kentucky, United States 2427.0 \n", "1104107 Shavano Park, TX 1497.0 \n", "1104108 Basking Ridge, NJ 1073.0 \n", "1104109 Austin, TX 1993.0 \n", "1104110 East Corinth, ME 718.0 \n", "\n", "[1104111 rows x 3 columns]" ] }, "execution_count": 41, "metadata": {}, "output_type": "execute_result" } ], "source": [ "%%time\n", "# Group by company, job_title, and month, and calculate the mean of summary_length\n", "grouped_df = df_merged.groupby(['job_title', 'job_location']).agg({'summary_length': 'mean'})\n", "\n", "# Reset index to sort by job_title and month\n", "grouped_df = grouped_df.reset_index()\n", "\n", "# Sort by job_title and month\n", "sorted_df = grouped_df.sort_values(by=['job_title', 'job_location','summary_length'],\n", " ascending=False).reset_index(drop=True).fillna(0)\n", "sorted_df" ] }, { "cell_type": "markdown", "id": "08c97b81-64c5-48fb-8fe0-d36789cf3deb", "metadata": { "id": "08c97b81-64c5-48fb-8fe0-d36789cf3deb" }, "source": [ "The acceleration is consistently 10x+ for complex aggregations and sorting that involve multiple columns." ] }, { "cell_type": "markdown", "id": "9bcc719b-666a-4bc9-97d6-16f448b5c707", "metadata": { "id": "9bcc719b-666a-4bc9-97d6-16f448b5c707" }, "source": [ "# Summary\n", "\n", "With cudf.pandas, you can keep using pandas as your primary dataframe library. When things start to get a little slow, just load the `cudf.pandas` extension and enjoy the incredible speedups.\n", "\n", "To learn more about cudf.pandas, we encourage you to visit https://rapids.ai/cudf-pandas." ] } ], "metadata": { "accelerator": "GPU", "colab": { "gpuType": "T4", "provenance": [] }, "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": "3.12.11" } }, "nbformat": 4, "nbformat_minor": 5 }