src/streamlit_app.py

import streamlit as st
import pandas as pd
import joblib
import matplotlib.pyplot as plt
import seaborn as sns
import numpy as np

# --- Load model and encoders ---
model = joblib.load("DB_model_IMAN.pkl")
cat_encoders = joblib.load("categorical_encoders.pkl")  # encoders for categorical features
num_encoders = joblib.load("numerical_scaler.pkl")       # StandardScaler או דומה
df = pd.DataFrame(pd.read_excel('/Users/shryqb/PycharmProjects/PythonProject/bachlor/some_running/iman_project/Rabies__Weather__War_Combined_1.4.25.xlsx'))


# --- Get unique values for dropdowns ---
animal_species_options = df['Animal Species'].dropna().unique().tolist()
rabies_species_options = df['Rabies Species'].dropna().unique().tolist()
settlement_options = df['Settlement'].dropna().unique().tolist()
region_weather_options = df['Region_Weather'].dropna().unique().tolist()


# --- Feature lists ---
feature_names = ['Year', 'Animal Species', 'Rabies Species', 'Settlement', 'x', 'y',
                 'Region_Weather', 'Avg Temperature', 'Monthly Precipitation (mm)',
                 'Rainy Days', 'War in Israel']



categorical_features = ['Year','Animal Species', 'Rabies Species', 'Settlement', 'Region_Weather','War in Israel']
numerical_features = [ 'x', 'y', 'Avg Temperature', 'Monthly Precipitation (mm)', 'Rainy Days']




######################################################################################################################################################
# setting

# --- Streamlit page config ---
st.set_page_config(page_title="Rabies Multi-Target Prediction", layout="wide", page_icon="🦠")

# --- Background style ---
st.markdown(
    """
    <style>
    .stApp {
        background: linear-gradient(to right, #e0ffe0, #fff0e0);
        color: #333;
    }
    h1, h2, h3 {
        color: #aa0000;
    }
    </style>
    """,
    unsafe_allow_html=True
)
######################################################################################################################################################

st.title("Rabies Multi-Target Prediction Dashboard")
st.header("Predict New Record")

# --- Collect inputs with selectbox ---
year = st.number_input("Year", min_value=1900, max_value=2030, value=2025)
animal_species = st.selectbox("Animal Species", animal_species_options)
rabies_species = st.selectbox("Rabies Species", rabies_species_options)
settlement = st.selectbox("Settlement", settlement_options)
x_coord = st.number_input("x coordinate", value=0.0)
y_coord = st.number_input("y coordinate", value=0.0)
region_weather = st.selectbox("Region Weather", region_weather_options)
avg_temp = st.number_input("Avg Temperature", value=25.0)
monthly_precip = st.number_input("Monthly Precipitation (mm)", value=50.0)
rainy_days = st.number_input("Rainy Days", value=5)
war_in_israel = st.selectbox("War in Israel", ["No", "Yes"])

# --- Create input DataFrame ---
input_df = pd.DataFrame([[
    year, animal_species, rabies_species, settlement, x_coord, y_coord,
    region_weather, avg_temp, monthly_precip, rainy_days, war_in_israel
]], columns=feature_names)

# --- Encode categorical features ---
for col in categorical_features:
    if col in cat_encoders:
        input_df[col] = cat_encoders[col].transform(input_df[col].astype(str))

# --- Scale numerical features ---
if num_encoders is not None:
    input_df[numerical_features] = num_encoders.transform(input_df[numerical_features])




st.subheader("Input Data")
st.dataframe(input_df)
input_df['War in Israel'] = input_df['War in Israel'].map({'Yes': 1, 'No': 0})
# --- Make prediction ---
if st.button("Predict"):
    try:
        prediction = model.predict(input_df)
        st.subheader("Predicted Values with Confidence")

        if hasattr(model, "predict_proba"):
            proba_list = model.predict_proba(input_df)
            target_names = ['Region', 'Month']

            for i, target in enumerate(target_names):
                probs = proba_list[i]
                top_idx = probs.argsort(axis=1)[:, -2:][:, ::-1]
                top_conf = probs[0, top_idx[0,0]] * 100
                top_label = cat_encoders[target].inverse_transform([top_idx[0,0]])[0]
                #second_conf = probs[0, top_idx[0,1]] * 100
                #second_label = cat_encoders[target].inverse_transform([top_idx[0,1]])[0]

                st.write(f"{target}: Predict: {top_label} - Confident: {top_conf:.2f} %")
                #st.write(f"{target} Second best: {second_label} - Confident: {second_conf:.2f}%")

        else:
            st.success(f"Prediction: {prediction}")

        # --- Display prediction probabilities per target ---
        if hasattr(model, "predict_proba"):
            proba_list = model.predict_proba(input_df)
            target_names = ['Region', 'Month']

            for i, target in enumerate(target_names):
                probs = proba_list[i][0]  # הסתברויות לכל הקטגוריות
                categories = cat_encoders[target].classes_  # שמות הקטגוריות המקוריים

                # להכין DataFrame
                df_probs = pd.DataFrame({
                    "Category": categories,
                    "Probability (%)": probs * 100
                }).sort_values(by="Probability (%)", ascending=False)

                st.subheader(f"{target} Prediction Probabilities")
                #st.dataframe(df_probs)  # מציג גם טבלה

                # --- גרף בר ---
                fig, ax = plt.subplots(figsize=(10, 5))
                colors = [
                    "#FF5733",  # כתום-אדום זוהר
                    "#33FF57",  # ירוק זוהר
                    "#3357FF",  # כחול זוהר
                    "#FF33A8",  # ורוד זוהר
                    "#F3FF33",  # צהוב זוהר
                    "#33FFF6",  # תכלת זוהר
                    "#FF8C33"  # כתום כהה-זוהר
                ]
                sns.barplot(x="Category", y="Probability (%)", data=df_probs, palette=colors, ax=ax)
                plt.xticks(rotation=45, ha='right')
                plt.ylim(0, 100)
                st.pyplot(fig)

    except Exception as e:
        st.error(f"Error during prediction: {e}")