app.py

import gradio as gr
import pandas as pd
import numpy as np
import joblib

# -----------------------------
# 1️⃣ Load model and data
# -----------------------------
MODEL_PATH = "linear_model.pkl"
EXCEL_PATH = "excel sheet of plant 2.xlsx"

linear_model = joblib.load(MODEL_PATH)
df = pd.read_excel(EXCEL_PATH)
df.columns = df.columns.str.strip()

strain_names = df['pea plant strain'].unique().tolist()
row_options = ["None, Enter Manually"] + [str(i) for i in range(len(df))]

# -----------------------------
# 2️⃣ Autofill function
# -----------------------------
def autofill_fields(row_index):
    if row_index == "None, Enter Manually":
        return [None]*11  # Dose, Soil N, P, K, pH + 6 optional yield metrics
    row = df.iloc[int(row_index)]
    return (
        row['Dose (g/pot)'], row['Soil N (ppm)'], row['Soil P (ppm)'],
        row['Soil K (ppm)'], row['pH'], row['Chlorophyll (SPAD)'],
        row['Shoot Length (cm)'], row['Root Length (cm)'], row['Shoot Wt (g)'],
        row['Root Wt (g)'], row['Yield (g/pot)']
    )

# -----------------------------
# 3️⃣ Prediction function
# -----------------------------
def predict_linear(strain, dose, soil_n, soil_p, soil_k, ph,
                   chlorophyll, shoot_len, root_len, shoot_wt, root_wt, yield_gp,
                   actual_yield=None):
    logs = []

    # Validate required inputs
    required = [dose, soil_n, soil_p, soil_k, ph]
    if any(v is None for v in required):
        logs.append("[DEBUG] Missing numeric inputs!")
        return "⚠️ Fill all inputs", "", "", "\n".join(logs)
    logs.append("[DEBUG] Inputs received.")

    # Prepare DataFrame
    X_input = pd.DataFrame([{
        'pea plant strain': strain,
        'Dose (g/pot)': dose,
        'Soil N (ppm)': soil_n,
        'Soil P (ppm)': soil_p,
        'Soil K (ppm)': soil_k,
        'pH': ph
    }])
    logs.append(f"[DEBUG] Input DataFrame:\n{X_input}")

    # Predict all 7 target metrics
    y_pred = linear_model.predict(X_input)[0]
    logs.append(f"[DEBUG] Predicted values:\n{y_pred}")

    # Optional actual_yield comparison
    if actual_yield is not None:
        try:
            actual_val = float(actual_yield)
            abs_error = np.abs(actual_val - y_pred[5])  # assuming Yield (g/pot) at index 5
            logs.append(f"[DEBUG] Actual Yield: {actual_val}")
            logs.append(f"[DEBUG] Absolute Error: {abs_error}")
            return (f"**Predicted Metrics:** {y_pred.round(2)}",
                    f"**Actual Yield:** {actual_val}",
                    f"**Absolute Error:** {abs_error:.2f}",
                    "\n".join(logs))
        except:
            logs.append("[DEBUG] Actual yield invalid.")

    return (f"**Predicted Metrics:** {y_pred.round(2)}",
            "Actual Yield: Not provided",
            "Absolute Error: N/A",
            "\n".join(logs))

# -----------------------------
# 4️⃣ Gradio Interface
# -----------------------------
with gr.Blocks(title="Linear Regression Plant Predictor") as demo:
    gr.Markdown("<h1 style='text-align:center'>Linear Regression — Plant Yield Predictor</h1>")

    with gr.Row():
        with gr.Column(scale=1):
            gr.Markdown("### Plant & Strain")
            strain_input = gr.Dropdown(strain_names, label="Select Strain", value=strain_names[0])
            row_selector = gr.Dropdown(row_options, label="Select Row", value="None, Enter Manually")

            gr.Markdown("### Input Parameters")
            dose = gr.Number(label="Dose (g/pot)")
            soil_n = gr.Number(label="Soil N (ppm)")
            soil_p = gr.Number(label="Soil P (ppm)")
            soil_k = gr.Number(label="Soil K (ppm)")
            ph = gr.Number(label="pH")

            gr.Markdown("### Optional Yield Metrics")
            chlorophyll = gr.Number(label="Chlorophyll (SPAD)")
            shoot_len = gr.Number(label="Shoot Length (cm)")
            root_len = gr.Number(label="Root Length (cm)")
            shoot_wt = gr.Number(label="Shoot Wt (g)")
            root_wt = gr.Number(label="Root Wt (g)")
            yield_gp = gr.Number(label="Yield (g/pot)")

            actual_yield = gr.Number(label="Actual Yield (g/pot)")
            predict_btn = gr.Button("Predict", variant="primary")

        with gr.Column(scale=1):
            gr.Markdown("### Inference Result")
            pred_box = gr.Markdown("Awaiting prediction...")
            actual_box = gr.Markdown("")
            abs_box = gr.Markdown("")
            log_box = gr.Textbox(label="Debug Logs", lines=15)

    # Autofill callback
    row_selector.change(fn=autofill_fields, inputs=[row_selector],
                        outputs=[dose, soil_n, soil_p, soil_k, ph,
                                 chlorophyll, shoot_len, root_len,
                                 shoot_wt, root_wt, yield_gp])

    # Prediction callback
    predict_btn.click(fn=predict_linear,
                      inputs=[strain_input, dose, soil_n, soil_p, soil_k, ph,
                              chlorophyll, shoot_len, root_len, shoot_wt, root_wt, yield_gp,
                              actual_yield],
                      outputs=[pred_box, actual_box, abs_box, log_box])

# -----------------------------
# 5️⃣ Launch
# -----------------------------
if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860)