import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestRegressor
from sklearn.metrics import mean_absolute_error, mean_absolute_percentage_error, r2_score
import matplotlib.pyplot as plt

# Load data
data = pd.read_csv('ETH-USD.csv')

# Convert 'Date' to datetime format
data['Date'] = pd.to_datetime(data['Date'])

# Set 'Date' as the index
data.set_index('Date', inplace=True)

# Features and target variable
features = ['Open', 'High', 'Low', 'Adj Close', 'Volume']

# Add lag features
for lag in [2, 7, 15, 30]:
    data[f'Close_Lag_{lag}'] = data['Close'].shift(lag)
    features.append(f'Close_Lag_{lag}')

# Add rolling averages
for window in [7, 15, 30]:
    data[f'Close_Rolling_Mean_{window}'] = data['Close'].rolling(window=window).mean()
    data[f'Close_Rolling_Std_{window}'] = data['Close'].rolling(window=window).std()
    features.extend([f'Close_Rolling_Mean_{window}', f'Close_Rolling_Std_{window}'])

# Shift the features one day into the future
data[features] = data[features].shift(1)

# Drop rows with NaN values introduced by lag, rolling features, and shifting
data.dropna(inplace=True)

# Target variable
target = 'Close'

# Train-test split
train_size = int(len(data) * 0.8)
train_data, test_data = data.iloc[:train_size], data.iloc[train_size:]

# Initialize and train a model (Random Forest Regressor, for example)
model = RandomForestRegressor(n_estimators=100, random_state=42)
model.fit(train_data[features], train_data[target])

# Make predictions on the test set
predictions = model.predict(test_data[features])

# Evaluate the model
mae = mean_absolute_error(test_data[target], predictions)
mape = mean_absolute_percentage_error(test_data[target], predictions) * 100
r2 = r2_score(test_data[target], predictions)

print(f'Mean Absolute Error (MAE): {mae}')
print(f'Mean Absolute Percentage Error (MAPE): {mape:.2f}%')
print(f'R-squared (R2) Score: {r2:.4f}')