import matplotlib
import numpy as np
from sklearn.svm import OneClassSVM
from data.loader_dc import DataLoader
import matplotlib.pyplot as plt


class OcSvm:
    def __init__(self):
        self._clf: "OneClassSVM" = OneClassSVM(nu=7.1e-3, gamma=1.6e-7)  # scale

    def fit(self, data: "np.ndarray"):
        self._clf.fit(data)
        res: "np.ndarray" = self._clf.predict(data)
        print(f"train detail: {res[res == 1].size}/{data.shape[0]}")

    def predict(self, data: "np.ndarray", which: "int"):
        res: "np.ndarray" = self._clf.predict(data)
        print(f"predict detail: {res[res == which].size}/{data.shape[0]}")


def main():
    np.random.seed(5)
    loader = DataLoader()
    cls = OcSvm()
    cls.fit(loader.have)
    cls.predict(loader.test, 1)
    cls.predict(loader.none, -1)


if __name__ == "__main__":
    matplotlib.use("TkAgg")
    plt.rcParams["font.sans-serif"] = ["SimHei"]  # 用来正常显示中文标签SimHei
    plt.rcParams["axes.unicode_minus"] = False
    main()