src/components/model_timeseries_risk.py

import numpy as np
import pandas as pd
import tensorflow as tf
import joblib
import requests
from io import BytesIO


# ----------------------------
# Load model and scaler from HF Hub
# ----------------------------
model_url = "https://huggingface.co/samithcs/timeseries_risk/tree/main/timeseries_risk/lstm_risk_model.keras"
scaler_url = "https://huggingface.co/samithcs/timeseries_risk/tree/main/timeseries_risk/scaler.joblib"

# Download model to a local temp file
model_path = "/tmp/lstm_risk_model.keras"
with requests.get(model_url, stream=True) as r:
    r.raise_for_status()
    with open(model_path, "wb") as f:
        for chunk in r.iter_content(chunk_size=8192):
            f.write(chunk)

# Load model
model = tf.keras.models.load_model(model_path)

# Load scaler
response = requests.get(scaler_url)
scaler = joblib.load(BytesIO(response.content))

# ----------------------------
# Load your CSV dataset (optional)
# ----------------------------
data_path = "https://huggingface.co/samithcs/risk_predictor/tree/main/supply_chain_disruptions_features.csv"  
df = pd.read_csv(data_path)

region_col = "Order City"
region_name = "Shanghai"

df_region = df[df[region_col] == region_name].copy()
if len(df_region) < 100:
    
    df_region = df.sample(200, random_state=42) if len(df) >= 200 else df

feature_cols = [
    "Days for shipping (real)", "Sales per customer", "Order Item Discount",
    "Order Item Product Price", "Order Item Quantity"
]
label_col = "Late_delivery_risk"
seq_length = 7

X_all = df_region[feature_cols].fillna(0).astype(float).values
y_all = df_region[label_col].fillna(0).astype(int).values

X_scaled = scaler.transform(X_all)

X_seq, y_seq = [], []
for i in range(len(X_scaled) - seq_length):
    X_seq.append(X_scaled[i:i+seq_length])
    y_seq.append(y_all[i+seq_length])
X_seq = np.array(X_seq)
y_seq = np.array(y_seq)


# ----------------------------
# Train/test split (optional)
# ----------------------------
test_size = int(0.2 * len(X_seq))
X_train, X_test = X_seq[:-test_size], X_seq[-test_size:]
y_train, y_test = y_seq[:-test_size], y_seq[-test_size:]

# ----------------------------
# Evaluate using loaded model
# ----------------------------
test_loss, test_acc = model.evaluate(X_test, y_test)


# ----------------------------
# Prediction function
# ----------------------------
def predict_risk_for_next_day(sequence, threshold=0.5):
    seq_scaled = scaler.transform(sequence)
    seq_window = np.expand_dims(seq_scaled, axis=0)
    pred_prob = model.predict(seq_window)[0][0]
    pred_label = int(pred_prob > threshold)
    
    return pred_prob, pred_label

# Example usage
if X_test.shape[0] > 0:
    predict_risk_for_next_day(X_test[0], threshold=0.5)