diff --git a/1.Table Extraction/Extract CSV files and Tables from Websites.ipynb b/1.Table Extraction/Extract CSV files and Tables from Websites.ipynb index dffc3c5..5e20f07 100644 --- a/1.Table Extraction/Extract CSV files and Tables from Websites.ipynb +++ b/1.Table Extraction/Extract CSV files and Tables from Websites.ipynb @@ -2,13 +2,18 @@ "cells": [ { "cell_type": "code", - "execution_count": 1, "id": "9bcec48b", - "metadata": {}, - "outputs": [], + "metadata": { + "ExecuteTime": { + "end_time": "2025-11-17T04:54:47.969482Z", + "start_time": "2025-11-17T04:54:47.678799Z" + } + }, "source": [ "import pandas as pd" - ] + ], + "outputs": [], + "execution_count": 6 }, { "cell_type": "markdown", @@ -28,23 +33,78 @@ }, { "cell_type": "code", - "execution_count": 2, "id": "62ad1716", - "metadata": {}, - "outputs": [], + "metadata": { + "ExecuteTime": { + "end_time": "2025-11-17T04:55:16.088685Z", + "start_time": "2025-11-17T04:55:14.726566Z" + } + }, "source": [ "# reading 1 csv file from the website\n", - "df_premier21 = pd.read_csv('https://www.football-data.co.uk/mmz4281/2122/E0.csv')" - ] + "df_premier25 = pd.read_csv('https://www.football-data.co.uk/mmz4281/2526/E0.csv')" + ], + "outputs": [], + "execution_count": 7 }, { "cell_type": "code", - "execution_count": 3, "id": "e0b6be7b", - "metadata": {}, + "metadata": { + "ExecuteTime": { + "end_time": "2025-11-17T04:55:23.155786Z", + "start_time": "2025-11-17T04:55:23.132682Z" + } + }, + "source": [ + "# showing dataframe\n", + "df_premier25" + ], "outputs": [ { "data": { + "text/plain": [ + " Div Date Time HomeTeam AwayTeam FTHG FTAG FTR HTHG \\\n", + "0 E0 15/08/2025 20:00 Liverpool Bournemouth 4 2 H 1 \n", + "1 E0 16/08/2025 12:30 Aston Villa Newcastle 0 0 D 0 \n", + "2 E0 16/08/2025 15:00 Brighton Fulham 1 1 D 0 \n", + "3 E0 16/08/2025 15:00 Sunderland West Ham 3 0 H 0 \n", + "4 E0 16/08/2025 15:00 Tottenham Burnley 3 0 H 1 \n", + ".. .. ... ... ... ... ... ... .. ... \n", + "105 E0 09/11/2025 14:00 Aston Villa Bournemouth 4 0 H 2 \n", + "106 E0 09/11/2025 14:00 Brentford Newcastle 3 1 H 0 \n", + "107 E0 09/11/2025 14:00 Crystal Palace Brighton 0 0 D 0 \n", + "108 E0 09/11/2025 14:00 Nott'm Forest Leeds 3 1 H 1 \n", + "109 E0 09/11/2025 16:30 Man City Liverpool 3 0 H 2 \n", + "\n", + " HTAG ... B365CAHH B365CAHA PCAHH PCAHA MaxCAHH MaxCAHA AvgCAHH \\\n", + "0 0 ... 2.03 1.78 2.07 1.85 2.03 1.88 1.94 \n", + "1 0 ... 2.05 1.80 2.02 1.89 2.06 1.80 1.95 \n", + "2 0 ... 1.83 2.03 1.93 2.00 1.84 2.03 1.80 \n", + "3 0 ... 1.95 1.90 1.97 1.95 1.95 1.94 1.86 \n", + "4 0 ... 1.98 1.88 1.99 1.93 1.98 1.91 1.88 \n", + ".. ... ... ... ... ... ... ... ... ... \n", + "105 0 ... 2.00 1.85 2.05 1.88 2.00 1.85 1.95 \n", + "106 1 ... 2.05 1.80 2.14 1.81 2.05 1.83 2.02 \n", + "107 0 ... 2.05 1.80 2.10 1.84 2.05 1.80 2.01 \n", + "108 1 ... 1.88 1.98 1.91 2.02 1.92 1.98 1.87 \n", + "109 0 ... 1.83 2.03 1.94 1.98 1.93 2.03 1.86 \n", + "\n", + " AvgCAHA BFECAHH BFECAHA \n", + "0 1.76 2.14 1.86 \n", + "1 1.74 2.14 1.86 \n", + "2 1.96 1.91 2.08 \n", + "3 1.78 2.02 1.97 \n", + "4 1.83 2.07 1.92 \n", + ".. ... ... ... \n", + "105 1.83 2.09 1.91 \n", + "106 1.76 2.19 1.83 \n", + "107 1.77 2.12 1.88 \n", + "108 1.90 1.97 2.01 \n", + "109 1.91 1.94 2.05 \n", + "\n", + "[110 rows x 132 columns]" + ], "text/html": [ "