import pandas as pd
import matplotlib.pyplot as plt
from sklearn.preprocessing import MinMaxScaler

# 设置字体为 Times New Roman（需确保系统已安装）
plt.rcParams['font.family'] = 'Times New Roman'

# 原始数据
datas = [
    [2.674, 0.814, 0.579, 5.69, 2.51],
    [2.543, 0.823, 0.584, 5.31, 2.56],
    [2.411, 0.837, 0.591, 4.87, 2.59],
    [2.174, 0.866, 0.595, 4.21, 2.66],
    [2.276, 0.861, 0.594, 4.47, 2.62],
    [2.310, 0.855, 0.593, 4.68, 2.59],
]
metrics = ["CER", "Emo2V.", "S-SIM", "DNSV", "AutoPCP"]
x_labels = [0, 100, 200, 500, 1000, 2000]

# 转为 DataFrame
df = pd.DataFrame(datas, columns=metrics)
df['Step'] = x_labels

# 归一化指标
scaler = MinMaxScaler()
df_scaled = df.copy()
df_scaled[metrics] = scaler.fit_transform(df[metrics])

# 绘图（提高分辨率 dpi=300）
plt.figure(figsize=(8, 4), dpi=300)

for metric in metrics:
    plt.plot(df_scaled["Step"], df_scaled[metric], label=metric, linewidth=2)

# plt.xlabel("Training Data Size (H)", fontsize=14)
# plt.ylabel("Normalized Score", fontsize=14)
plt.xticks(x_labels, fontsize=16)
plt.yticks(fontsize=16)
plt.legend(fontsize=16)
plt.grid(True)
plt.tight_layout()

# 保存图像（可选）
plt.savefig("examples/celsds/infer/evaluate/normalized_metric_trends.png", dpi=300)

plt.show()