{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "WB9guq4AguTq"
      },
      "outputs": [],
      "source": [
        "\n",
        "# Install required packages\n",
        "\n",
        "!pip install --quiet yfinance joblib pmdarima huggingface_hub\n",
        "\n",
        "#  Import Libraries\n",
        "\n",
        "from huggingface_hub import hf_hub_download\n",
        "import joblib\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import yfinance as yf\n",
        "\n",
        "#  Hugging Face Token\n",
        "\n",
        "HF_TOKEN = \"your_own_hf_token\"   # Replace with your token\n",
        "\n",
        "\n",
        "# Load ARIMA Model & Box-Cox Transformer\n",
        "\n",
        "arima_model_path = hf_hub_download(\n",
        "    repo_id=\"EsferSami/DataSynthis_ML_JobTask\",\n",
        "    filename=\"Apple-Stock-Price-Forecasting-ARIMA-Model/apple_stock_arima.pkl\",\n",
        "    token=HF_TOKEN\n",
        ")\n",
        "\n",
        "bct_path = hf_hub_download(\n",
        "    repo_id=\"EsferSami/DataSynthis_ML_JobTask\",\n",
        "    filename=\"Apple-Stock-Price-Forecasting-ARIMA-Model/boxcox_transformer.pkl\",\n",
        "    token=HF_TOKEN\n",
        ")\n",
        "\n",
        "arima_model = joblib.load(arima_model_path)\n",
        "bct = joblib.load(bct_path)\n",
        "\n",
        "print(\"ARIMA model  loaded from Hugging Face Hub!\")\n",
        "\n",
        "\n",
        "# Download Recent AAPL Data\n",
        "\n",
        "data = yf.download(\"AAPL\", period=\"3mo\", auto_adjust=False)\n",
        "recent_prices = data['Adj Close'].values.astype(float)\n",
        "\n",
        "\n",
        "\n",
        "# Transform Data using Box-Cox\n",
        "\n",
        "y_trans, _ = bct.transform(recent_prices)\n",
        "\n",
        "\n",
        "# Compute residuals standard deviation for small fluctuations\n",
        "\n",
        "try:\n",
        "    residuals = arima_model.resid()\n",
        "    resid_std = np.std(residuals)\n",
        "except Exception:\n",
        "    resid_std = np.std(y_trans - np.mean(y_trans))  # fallback\n",
        "print(f\"Residual std used for fluctuation: {resid_std:.4f}\")\n",
        "\n",
        "\n",
        "# 7-Day Forecast with Small Fluctuations\n",
        "\n",
        "predictions_trans = []\n",
        "current_series = y_trans.copy()\n",
        "\n",
        "for day in range(7):\n",
        "    # Predict next step using ARIMA\n",
        "    try:\n",
        "        pred = arima_model.predict(n_periods=1)[0]\n",
        "    except Exception:\n",
        "        pred = current_series[-1]\n",
        "\n",
        "    # Anchor to last actual price in Box-Cox space\n",
        "    last_price_bc = current_series[-1]\n",
        "\n",
        "    # Add small random fluctuation (±30% of residual std)\n",
        "    pred = last_price_bc + np.random.normal(loc=0.0, scale=resid_std * 0.3)\n",
        "\n",
        "    predictions_trans.append(pred)\n",
        "    current_series = np.append(current_series, pred)\n",
        "\n",
        "\n",
        "# Inverse Box-Cox to Price Space\n",
        "\n",
        "predictions_price, _ = bct.inverse_transform(np.array(predictions_trans))\n",
        "\n",
        "\n",
        "#  Prepare Results Table\n",
        "\n",
        "prediction_dates = pd.date_range(start=data.index[-1] + pd.Timedelta(days=1), periods=7)\n",
        "arima_results_df = pd.DataFrame({\n",
        "    'Date': prediction_dates,\n",
        "    'Predicted_Price': predictions_price\n",
        "})\n",
        "\n",
        "print(\"\\nARIMA - 7-Day Forecast\")\n",
        "print(\"=\"*60)\n",
        "print(arima_results_df.to_string(index=False))\n"
      ]
    }
  ]
}