Reproducibility capsule - init, exp 1 plots, v1, not yet final (#5)

* figure comparison experiment 1

* figure comparison experiment 1 - v1

* figure comparison experiment 1 - v2

* exp 1 cpu utilization, latency, with mock data - v1

* exp 1 cpu utilization, latency, with mock data - v1.1

Author: Radu-Nicolae

reproducibility_capsule/exp_1/data/footprinter.parquet

@@ -0,0 +1,147 @@
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+from matplotlib.ticker import FuncFormatter
+import os
+
+# --- 1. CONFIGURATION & STYLE ---
+COLOR_PALETTE = [
+    "#0072B2", "#E69F00", "#009E73", "#D55E00", "#CC79A7", "#F0E442", "#8B4513",
+    "#56B4E9", "#F0A3FF", "#FFB400", "#00BFFF", "#90EE90", "#FF6347", "#8A2BE2",
+    "#CD5C5C", "#4682B4", "#FFDEAD", "#32CD32", "#D3D3D3", "#999999"
+]
+METRIC = "power_draw"
+
+
+def load_and_process_data():
+    """Loads parquet files and aligns timestamps."""
+    print("Loading data...")
+
+    # Check if data exists (Warning only)
+    if not os.path.exists("../data/footprinter.parquet"):
+        print("Warning: ../data/footprinter.parquet not found.")
+
+    # Load Data
+    fp = pd.read_parquet("../data/footprinter.parquet").groupby("timestamp")[METRIC].sum()
+    odt = pd.read_parquet("../data/opendt.parquet").groupby("timestamp")[METRIC].sum()
+    rw = pd.read_parquet("../data/real_world.parquet").groupby("timestamp")[METRIC].sum()
+
+    # --- Processing ---
+    print("Processing and aligning data...")
+
+    def average_every_n(series, n):
+        return series.groupby(np.arange(len(series)) // n).mean()
+
+    # Average to 5-min intervals
+    # OpenDT (2.5m) -> 2 samples = 5 min
+    # Others (30s) -> 10 samples = 5 min
+    odt = average_every_n(odt, 2)
+    fp = average_every_n(fp, 10)
+    rw = average_every_n(rw, 10)
+
+    # Sync lengths (trim to shortest)
+    min_len = min(len(odt), len(fp), len(rw))
+    odt = odt.iloc[:min_len]
+    fp = fp.iloc[:min_len]
+    rw = rw.iloc[:min_len]
+
+    # Force Start Time to 2022-10-06 22:00:00
+    start_time = pd.Timestamp("2022-10-06 22:00:00")
+    # Fixed deprecation warning: used '5min' instead of '5T'
+    timestamps = pd.date_range(start=start_time, periods=min_len, freq="5min")
+
+    # Apply clean timestamps
+    odt.index = timestamps
+    fp.index = timestamps
+    rw.index = timestamps
+
+    return fp, odt, rw, timestamps, min_len
+
+
+def calculate_mape(ground_truth, simulation):
+    """Calculates Mean Absolute Percentage Error (MAPE)."""
+    R = ground_truth.values
+    S = simulation.values
+    return np.mean(np.abs((R - S) / R)) * 100
+
+
+def generate_experiment_pdf(x, fp, odt, rw, timestamps, min_len):
+    """Generates the final publication plot."""
+    print("Generating final experiment PDF...")
+
+    # Setup Figure (12, 5 size)
+    plt.figure(figsize=(12, 5))
+    plt.grid(True)
+
+    # Plot Lines (Thick lines: lw=3)
+    plt.plot(x, rw.values / 1000, label="Ground Truth", color=COLOR_PALETTE[0], lw=3)
+    plt.plot(x, fp.values / 1000, label="FootPrinter", color=COLOR_PALETTE[1], lw=3)
+    plt.plot(x, odt.values / 1000, label="OpenDT", color=COLOR_PALETTE[2], lw=3)
+
+    ax = plt.gca()
+
+    # --- Formatting X-Axis (Fixed Dates) ---
+    target_dates = ["2022-10-08", "2022-10-10", "2022-10-12", "2022-10-14"]
+    tick_dates = pd.to_datetime(target_dates)
+
+    tick_positions = []
+    tick_labels = []
+
+    for d in tick_dates:
+        seconds_diff = (d - timestamps[0]).total_seconds()
+        # 300 seconds = 5 minutes
+        idx = int(seconds_diff / 300)
+
+        tick_positions.append(idx)
+        tick_labels.append(d.strftime("%d/%m"))
+
+    ax.set_xticks(tick_positions)
+    # INCREASED FONT SIZE (was 14)
+    ax.set_xticklabels(tick_labels, fontsize=20)
+
+    # Extend limit slightly to show last tick
+    max_tick = max(tick_positions)
+    if ax.get_xlim()[1] < max_tick:
+        ax.set_xlim(right=max_tick + (min_len * 0.02))
+
+    # --- Formatting Y-Axis ---
+    y_formatter = FuncFormatter(lambda val, _: f"{int(val):,}")
+    ax.yaxis.set_major_formatter(y_formatter)
+    # INCREASED FONT SIZE (was 14)
+    ax.tick_params(axis='y', labelsize=20)
+
+    # Labels (INCREASED FONT SIZE)
+    plt.ylabel("Power Draw [kW]", fontsize=22, labelpad=10)
+    plt.xlabel("Time [day/month]", fontsize=22, labelpad=10)
+    plt.ylim(bottom=0)
+
+    # Legend
+    # FIXED: Cleaned 'loc' and ensured it matches bbox placement
+    plt.legend(fontsize=18, loc="upper center", bbox_to_anchor=(0.5, 1.22), ncol=3, framealpha=1)
+
+    plt.tight_layout()
+
+    # Save
+    plt.savefig("exp1_plot_power_draw.pdf", format="pdf", bbox_inches="tight")
+    print("Plot saved as 'exp1_plot_power_draw.pdf'")
+    # plt.show() # Uncomment if running in interactive mode/notebook
+    plt.close()
+
+
+def main():
+    # 1. Load Data
+    fp, odt, rw, timestamps, min_len = load_and_process_data()
+    x = np.arange(min_len)
+
+    # 2. Calculate Stats
+    mape_fp = calculate_mape(rw, fp)
+    mape_odt = calculate_mape(rw, odt)
+    print(f"Stats Calculated (Not plotted) - FP: {mape_fp:.2f}%, ODT: {mape_odt:.2f}%")
+
+    # 3. Generate Plot
+    generate_experiment_pdf(x, fp, odt, rw, timestamps, min_len)
+    print("Done!")
+
+
+if __name__ == "__main__":
+    main()