import seaborn as sns import streamlit as st from st_aggrid import AgGrid, GridOptionsBuilder, GridUpdateMode import requests import polars as pl from datetime import date import pandas as pd import matplotlib # Display the app title and description st.markdown(""" ## tjStuff+ App ##### By: Thomas Nestico ([@TJStats](https://x.com/TJStats)) ##### Code: [GitHub Repo](https://github.com/tnestico/streamlit_tjstuff) ##### Data: [MLB](https://baseballsavant.mlb.com/) ([Gathered from my MLB Scraper](https://github.com/tnestico/mlb_scraper)) #### About This Streamlit app tabulates and plots my pitching metric, tjStuff+, for all MLB players during the 2024 MLB Season About tjStuff+: * tjStuff+ calculates the Expected Run Value (xRV) of a pitch regardless of type * tjStuff+ is normally distributed, where 100 is the mean and Standard Deviation is 10 * Pitch Grade is based off tjStuff+ and scales the data to the traditional 20-80 Scouting Scale for a given pitch type [Learn More about tjStuff+ here](https://github.com/tnestico/tjstuff_plus/tree/main) """ ) # Dictionary to map pitch types to their corresponding colors and names pitch_colours = { ## Fastballs ## 'FF': {'colour': '#FF007D', 'name': '4-Seam Fastball'}, 'FA': {'colour': '#FF007D', 'name': 'Fastball'}, 'SI': {'colour': '#98165D', 'name': 'Sinker'}, 'FC': {'colour': '#BE5FA0', 'name': 'Cutter'}, ## Offspeed ## 'CH': {'colour': '#F79E70', 'name': 'Changeup'}, 'FS': {'colour': '#FE6100', 'name': 'Splitter'}, 'SC': {'colour': '#F08223', 'name': 'Screwball'}, 'FO': {'colour': '#FFB000', 'name': 'Forkball'}, ## Sliders ## 'SL': {'colour': '#67E18D', 'name': 'Slider'}, 'ST': {'colour': '#1BB999', 'name': 'Sweeper'}, 'SV': {'colour': '#376748', 'name': 'Slurve'}, ## Curveballs ## 'KC': {'colour': '#311D8B', 'name': 'Knuckle Curve'}, 'CU': {'colour': '#3025CE', 'name': 'Curveball'}, 'CS': {'colour': '#274BFC', 'name': 'Slow Curve'}, 'EP': {'colour': '#648FFF', 'name': 'Eephus'}, ## Others ## 'KN': {'colour': '#867A08', 'name': 'Knuckleball'}, 'PO': {'colour': '#472C30', 'name': 'Pitch Out'}, 'UN': {'colour': '#9C8975', 'name': 'Unknown'}, } # Create dictionaries for pitch types and their attributes dict_colour = {key: value['colour'] for key, value in pitch_colours.items()} dict_pitch = {key: value['name'] for key, value in pitch_colours.items()} dict_pitch_desc_type = {value['name']: key for key, value in pitch_colours.items()} dict_pitch_name = {value['name']: value['colour'] for key, value in pitch_colours.items()} # Define a custom colormap for styling cmap_sum = matplotlib.colors.LinearSegmentedColormap.from_list("", ['#648FFF', '#FFFFFF', '#FFB000']) # Initialize session state for cache status if 'cache_cleared' not in st.session_state: st.session_state.cache_cleared = False # Function to fetch data and cache it @st.cache_data def fetch_data(): df = pl.read_csv("tjstuff_plus_pitch_data_2024.csv").fill_nan(None) return df # Fetch and preprocess data df = fetch_data() df_plot = df.clone() df = df.filter(df['pitches'] >= 10).drop_nulls(subset=['pitch_grade', 'tj_stuff_plus']) df = df.sort(['pitcher_name', 'pitch_type'], descending=[False, False]) # Cast columns to appropriate data types df = df.with_columns([ pl.col('tj_stuff_plus').cast(pl.Int64).alias('tj_stuff_plus'), pl.col('pitches').cast(pl.Int64).alias('pitches'), pl.col('pitcher_id').cast(pl.Int64).alias('pitcher_id'), pl.col('pitch_grade').cast(pl.Int64).alias('pitch_grade') ]) # Define column configuration for Streamlit column_config_dict = { 'pitcher_id': 'Pitcher ID', 'pitcher_name': 'Pitcher Name', 'pitch_type': 'Pitch Type', 'pitches': 'Pitches', 'tj_stuff_plus': st.column_config.NumberColumn("tjStuff+", format="%.0f"), 'pitch_grade': st.column_config.NumberColumn("Pitch Grade", format="%.0f") } # Get unique pitch types for selection unique_pitch_types = [''] + sorted(df['pitch_type'].unique().to_list()) unique_pitch_types = [dict_pitch.get(x, x) for x in unique_pitch_types] st.markdown(""" #### tjStuff+ Table Filter and sort tjStuff+ Data for all MLB Pitchers """ ) # Create a selectbox widget for pitch types selected_pitch_types = st.selectbox('Select Pitch Types *(leave blank for all pitch types)*', unique_pitch_types) # Create a selectbox widget for position selected_position = st.selectbox('Select Position *(leave blank for all Pitchers)*', ['','SP','RP']) # Filter the DataFrame based on selected pitch types if selected_pitch_types == 'All': df = df.filter(pl.col('pitch_type') == 'All').sort('tj_stuff_plus', descending=True) elif selected_pitch_types != '': df = df.filter(pl.col('pitch_type') == dict_pitch_desc_type[selected_pitch_types]).sort('tj_stuff_plus', descending=True) if selected_position != '': df = df.filter(pl.col('position') == selected_position).sort('tj_stuff_plus', descending=True) # Convert Polars DataFrame to Pandas DataFrame and apply styling styled_df = df[['pitcher_id', 'pitcher_name', 'pitch_type', 'pitches', 'tj_stuff_plus', 'pitch_grade']].to_pandas().style # Apply background gradient styling to specific columns styled_df = styled_df.background_gradient(subset=['tj_stuff_plus'], cmap=cmap_sum, vmin=80, vmax=120) styled_df = styled_df.background_gradient(subset=['pitch_grade'], cmap=cmap_sum, vmin=20, vmax=80) # Display the styled DataFrame in Streamlit st.dataframe(styled_df, hide_index=True, column_config=column_config_dict, width=1500) # Create dictionaries for pitcher information pitcher_id_name = dict(zip(df_plot['pitcher_id'], df_plot['pitcher_name'])) pitcher_id_name_id = dict(zip(df_plot['pitcher_id'], df_plot['pitcher_name'] + ' - ' + df_plot['pitcher_id'])) pitcher_name_id_id = dict(zip(df_plot['pitcher_name'] + ' - ' + df_plot['pitcher_id'], df_plot['pitcher_id'])) pitcher_id_position = dict(zip(df_plot['pitcher_id'], df_plot.drop_nulls(subset=['position'])['position'])) st.markdown(""" #### tjStuff+ Plot Visualize tjStuff+ and Pitching Grade by Pitcher """ ) # Create a selectbox widget for pitchers pitcher_id_name_select = st.selectbox('Select Pitcher', sorted(pitcher_name_id_id.keys())) # Get selected pitcher information pitcher_id = pitcher_name_id_id[pitcher_id_name_select] position = pitcher_id_position[pitcher_id] pitcher_name = pitcher_id_name[pitcher_id] import tjstuff_plot # Button to update plot # Get selected pitcher information pitcher_id = pitcher_name_id_id[pitcher_id_name_select] position = pitcher_id_position[pitcher_id] pitcher_name = pitcher_id_name[pitcher_id] import tjstuff_plot # Button to update plot if st.button('Update Plot'): st.session_state.update_plot = True tjstuff_plot.tjstuff_plot(df_plot, pitcher_id, position, pitcher_name)