explain

app.py

@@ -105,14 +105,6 @@ def like(issue):
 '''
 EXPLAINABLE AI
 '''
-import shap
-df=pd.read_csv("xgb/re.csv")
-df.columns=['id', 'date', 'age', 'dist_subway', 'dist_stores', 'lat', 'long', 'price']
-data=df.loc[:,['age', 'dist_subway', 'dist_stores', 'lat', 'long', 'price']]
-
-from sklearn.model_selection import train_test_split
-X,y=data.iloc[:,:-1],data.iloc[:,-1]
-X_train, X_test, y_train,y_test=train_test_split(X,y, test_size=0.1, random_state=42)
 
 
 '''
@@ -228,25 +220,30 @@ With no need for jargon, SSDS delivers tangible value to our fintech operations.
       """)
 
   with gr.Tab("Explainable AI"):
+    gr.Markdown("""
+    Explain by Dataset
+    =============
+    ![](xgb/data.png)
+    sorted feature from top(most import)
+    
+    dist_subway when at low value(green) make big impact to price
+    
+    Explain by Feature
+    =============
+    ![](xgb/feature.png)
+    dist lower than 900 spike the price f(x)
+    
+    also highlighted the shap value for record[20] at around 6500
+    
+    Explain by Record
+    =============
+    ![](xgb/record.png)
+    the largest contribution to positive price is dist_subway
+    
+    second contribution is age
 
+    """)
 
-    file_name = 'xgb/xgb.model'
-    bst = pickle.load(open(file_name, "rb"))
-    explainer_xgb = shap.Explainer(bst)  # , X100)
-    shap_values_xgb = explainer_xgb(X_test)
-
-    sample_ind = 20
-    shap.partial_dependence_plot(
-      "dist_subway",
-      bst.predict,
-      X_test,
-      model_expected_value=True,
-      feature_expected_value=True,
-      ice=False,
-      shap_values=shap_values_xgb[sample_ind: sample_ind + 1, :],
-    )
-    shap.force_plot(explainer_xgb.expected_value, shap_values_xgb[0].values, X_test.iloc[0], matplotlib=True)
-    shap.plots.beeswarm(shap_values_xgb)
 
   with gr.Tab("Fine Tune LLM"):
     in_like = gr.Textbox(placeholder="having credit card problem"   ,                                             label="Issue",

requirements.txt

@@ -5,5 +5,3 @@ pandas==2.1.3
 #yfinance==0.2.31
 #scikit-learn
 plotly
-shap
-xgboost