TransferIQ - Final Submission

README.md

@@ -1,2 +1,11 @@
-# Transfer-IQ
-Dynamic Player Transfer Value Prediction using AI and Multi-source Data
+# TransferIQ
+
+Dynamic Player Transfer Value Prediction using AI and Multi-source Data.
+
+TransferIQ is an AI-powered football analytics dashboard that predicts player market value using machine learning models such as LSTM and XGBoost.
+
+The system analyzes player performance, injuries, and sentiment to generate accurate predictions and insights. It provides interactive visualizations like market trends, performance radar, and model comparisons using Streamlit and Plotly.
+
+This project helps in making data-driven transfer decisions by combining analytics and predictive modeling in a user-friendly dashboard.
+
+🚀 Live Demo: https://transferiq-dashboard.onrender.com/

requirements.txt

@@ -0,0 +1,7 @@
+streamlit
+pandas
+numpy
+scikit-learn
+xgboost
+plotly
+statsmodels

transferiq_dashboard.py

@@ -0,0 +1,214 @@
+import streamlit as st
+import pandas as pd
+import plotly.express as px
+import plotly.graph_objects as go
+import numpy as np
+from sklearn.metrics import mean_squared_error
+
+st.set_page_config(page_title="TransferIQ Player Analytics", layout="wide")
+
+# -------------------------------
+# LOAD DATA
+# -------------------------------
+df = pd.read_csv("final_feature_dataset_with_predictions.csv", encoding="latin1")
+
+st.title("⚽ TransferIQ Dashboard")
+st.markdown("### 📊 AI-Powered Football Player Value Analysis")
+
+# -------------------------------
+# OVERVIEW
+# -------------------------------
+st.markdown("## 📌 Overview")
+
+col1, col2, col3 = st.columns(3)
+col1.metric("Total Players", len(df["player_name"].unique()))
+col2.metric("Seasons", len(df["season"].unique()))
+col3.metric("Model", "LSTM + XGBoost")
+
+st.divider()
+
+# -------------------------------
+# PLAYER SELECTION
+# -------------------------------
+st.markdown("## 👤 Player Analysis")
+
+players = df["player_name"].unique()
+player = st.selectbox("Select Player", players)
+
+years = sorted(df["season"].unique())
+year = st.selectbox("Select Season", years)
+
+player_data = df[(df["player_name"] == player) & (df["season"] == year)]
+full_player_data = df[df["player_name"] == player]
+
+# -------------------------------
+# PROFILE
+# -------------------------------
+st.subheader("Player Profile")
+
+if len(player_data) > 0:
+    c1, c2, c3 = st.columns(3)
+    c1.metric("Position", player_data["position"].iloc[0])
+    c2.metric("Market Value", f"€{int(player_data['market_value_eur'].iloc[0]):,}")
+    c3.metric("Availability", round(player_data["availability_rate"].iloc[0], 2))
+
+st.divider()
+
+# -------------------------------
+# MARKET TREND
+# -------------------------------
+st.subheader("📈 Market Value Trend")
+fig = px.line(full_player_data, x="season", y="market_value_eur", markers=True)
+st.plotly_chart(fig, width="stretch")
+
+# -------------------------------
+# RADAR
+# -------------------------------
+st.subheader("⚽ Performance Radar")
+
+metrics = ["goals_per90","assists_per90","shots_per90","dribbles_per90","defensive_actions_per90"]
+
+if len(player_data) > 0:
+    fig_radar = go.Figure()
+    fig_radar.add_trace(go.Scatterpolar(
+        r=player_data[metrics].iloc[0].values,
+        theta=metrics,
+        fill='toself'
+    ))
+    st.plotly_chart(fig_radar, width="stretch")
+
+# -------------------------------
+# INJURY
+# -------------------------------
+st.subheader("🏥 Injury Analysis")
+
+if len(player_data) > 0:
+    injury_df = pd.DataFrame({
+        "Metric":["Injuries","Days","Matches Missed"],
+        "Value":[
+            player_data["total_injuries"].iloc[0],
+            player_data["total_days_injured"].iloc[0],
+            player_data["total_matches_missed"].iloc[0]
+        ]
+    })
+    fig2 = px.bar(injury_df, x="Metric", y="Value", color="Metric")
+    st.plotly_chart(fig2, width="stretch")
+
+# -------------------------------
+# SENTIMENT
+# -------------------------------
+st.subheader("💬 Sentiment Score")
+
+if len(player_data) > 0:
+    s = player_data["overall_sentiment"].iloc[0]
+    st.progress(float((s+1)/2))
+    st.write("Score:", round(s,3))
+
+st.divider()
+
+# -------------------------------
+# COMPARISON
+# -------------------------------
+st.subheader("🔍 Player Comparison")
+
+player2 = st.selectbox("Compare with", players)
+player2_data = df[df["player_name"] == player2]
+
+comp_df = pd.DataFrame({
+    "Metric":metrics,
+    player:full_player_data[metrics].iloc[0].values,
+    player2:player2_data[metrics].iloc[0].values
+})
+
+fig_compare = px.bar(comp_df, x="Metric", y=[player,player2], barmode="group")
+st.plotly_chart(fig_compare, width="stretch")
+
+# ======================================================
+# MODEL ANALYSIS
+# ======================================================
+st.divider()
+st.markdown("## 🤖 Model Performance")
+
+valid_df = df.dropna(subset=["lstm_pred","xgb_pred","final_pred"]).copy()
+valid_df = valid_df.sort_values(by="season")
+
+rmse_lstm = np.sqrt(mean_squared_error(valid_df["market_value_eur"], valid_df["lstm_pred"]))
+rmse_xgb = np.sqrt(mean_squared_error(valid_df["market_value_eur"], valid_df["xgb_pred"]))
+rmse_final = np.sqrt(mean_squared_error(valid_df["market_value_eur"], valid_df["final_pred"]))
+
+c1, c2, c3 = st.columns(3)
+c1.metric("LSTM RMSE", round(rmse_lstm,2))
+c2.metric("XGBoost RMSE", round(rmse_xgb,2))
+c3.metric("Ensemble RMSE", round(rmse_final,2))
+
+fig_model = px.bar(
+    x=["LSTM","XGBoost","Ensemble"],
+    y=[rmse_lstm, rmse_xgb, rmse_final],
+    text=[round(rmse_lstm,2),round(rmse_xgb,2),round(rmse_final,2)]
+)
+fig_model.update_traces(textposition="outside")
+st.plotly_chart(fig_model, width="stretch")
+
+best_model = min(
+    {"LSTM": rmse_lstm, "XGBoost": rmse_xgb, "Ensemble": rmse_final},
+    key=lambda x: {"LSTM": rmse_lstm, "XGBoost": rmse_xgb, "Ensemble": rmse_final}[x]
+)
+
+st.success(f"🏆 Best Model: {best_model}")
+
+# ======================================================
+# FINAL PREDICTIONS
+# ======================================================
+st.divider()
+st.markdown("## 🔮 Final Predictions")
+
+results_df = valid_df[[
+    "player_name",
+    "market_value_eur",
+    "lstm_pred",
+    "xgb_pred",
+    "final_pred"
+]].copy()
+
+results_df.columns = ["Player","Actual","LSTM","XGBoost","Final"]
+
+st.dataframe(results_df.head(10), width="stretch")
+
+st.subheader("🌟 Top Players")
+st.dataframe(results_df.sort_values("Final", ascending=False).head(5))
+
+fig_final = go.Figure()
+fig_final.add_trace(go.Scatter(
+    x=valid_df["season"],
+    y=valid_df["market_value_eur"],
+    name="Actual",
+    mode="lines+markers"
+))
+fig_final.add_trace(go.Scatter(
+    x=valid_df["season"],
+    y=valid_df["final_pred"],
+    name="Predicted",
+    mode="lines+markers"
+))
+st.plotly_chart(fig_final, width="stretch")
+
+# ======================================================
+# DOWNLOAD (FIXED 🔥)
+# ======================================================
+st.download_button(
+    "⬇ Download Predictions",
+    valid_df.to_csv(index=False),
+    "predictions.csv"
+)
+
+# ======================================================
+# INSIGHTS
+# ======================================================
+st.markdown("## 💡 Insights")
+
+st.info("""
+✔ Performance strongly affects value  
+✔ Sentiment improves prediction  
+✔ Ensemble model gives best stability  
+✔ Useful for transfer decision making  
+""")