mutis.signal.Signal.OU_fit()

Complexity

Conditions

8

Size

Total Lines	101
Code Lines	55

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

# Licensed under a 3-clause BSD style license - see LICENSE
    def OU_fit(self, bins=None, rang=None, a=1e-5, b=100):

Method name "OU_fit" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

Argument name "a" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

Argument name "b" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

This function exceeds the maximum number of variables (20/15).

        """Fit the signal to an OU stochastic process, using several statistical approaches.

        This function tries to fit the signal to an OU stochastic
        process using both basic curve fitting and the Maximum
        Likelihood Estimation method, and returns some plots of
        the signals and its properties, and the estimated parameters.
        """

        # TODO: make a generic fit method

TODO and FIXME comments should generally be avoided.

        res = dict()

        y = self.values

Variable name "y" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

        t = self.times

Variable name "t" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

        dy = np.diff(y)

Variable name "dy" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

        dt = np.diff(t)

Variable name "dt" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

        # estimate sigma
        try:
            sigma_est = (np.nanmean(dy ** 2 / y[:-1] ** 2 / dt)) ** 0.5
            res["sigma_est"] = sigma_est
        except Exception as e:

Catching very general exceptions such as Exception is usually not recommended.

Variable name "e" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

            log.error(f"Could not estimate sigma: {e}")

Use lazy % formatting in logging functions

        # plot histogram
        if bins is None:
            bins = int(y.size ** 0.5 / 1.5)  # bins='auto'
        if rang is None:
            rang = (np.percentile(y, 0), np.percentile(y, 99))

        p, x = np.histogram(y, density=True, bins=bins, range=rang)  # bins='sqrt')

Variable name "p" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

Variable name "x" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

        x = (x + np.roll(x, -1))[:-1] / 2.0

Variable name "x" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

        plt.subplots()

        plt.hist(y, density=True, alpha=0.75, bins=bins, range=rang)
        plt.plot(x, p, "r-", alpha=0.5)

        anchored_text = AnchoredText(
            f"mean    {np.mean(y):.2g} \n "
            f"median  {np.median(y):.2g} \n "
            f"mode    {scipy_stats.mode(y)[0]:.2g} \n "
            f"std     {np.std(y):.2g} \n "
            f"var     {np.var(y):.2g}",
            loc="upper right",
        )
        plt.gca().add_artist(anchored_text)

        # curve_fit
        try:
            popt, pcov = scipy_optimize.curve_fit(f=self.pdf, xdata=x, ydata=p)
            # print('curve_fit: (l, mu)')
            # print('popt: ')
            # print(popt)
            # print('pcov: ')
            # print(np.sqrt(np.diag(pcov)))
            x_c = np.linspace(1e-5, 1.1 * np.max(x), 1000)
            plt.plot(x_c, self.pdf(x_c, *popt), "k-", label="curve_fit", alpha=0.8)
            res["curve_fit"] = (popt, np.sqrt(np.diag(pcov)))
        except Exception as e:

Catching very general exceptions such as Exception is usually not recommended.

Variable name "e" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

            log.error(f"Some error fitting with curve_fit {e}")

Use lazy % formatting in logging functions

        # fit pdf with MLE
        # TODO: place outside as a helper class

TODO and FIXME comments should generally be avoided.

        #       mu here is different than self.mu
        class OU(scipy_stats.rv_continuous):

Missing class docstring

            def _pdf(self, x, l, mu):

Argument name "x" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

Argument name "l" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

Argument name "mu" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

This method could be written as a function/class method.

                return (
                    (l * mu) ** (1 + l)
                    / scipy_special.gamma(1 + l)
                    * np.exp(-l * mu / x)
                    / x ** (l + 2)
                )

        try:
            fit = OU(a=a, b=np.percentile(y, b)).fit(y, 1, 1, floc=0, fscale=1)
            # print('MLE fit: (l, mu)')
            # print(fit)
            x_c = np.linspace(0, 1.1 * np.max(x), 1000)
            plt.plot(x_c, self.pdf(x_c, fit[0], fit[1]), "k-.", label="MLE", alpha=0.8)
            res["MLE_fit"] = fit[:-2]
        except Exception as e:

Catching very general exceptions such as Exception is usually not recommended.

Variable name "e" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

            log.error(f"Some error fitting with MLE {e}")

Use lazy % formatting in logging functions

        plt.legend(loc="lower right")
        # plt.show()

        # estimate theta (from curve_fit)
        try:
            res["th_est1"] = fit[0] * sigma_est ** 2 / 2
        except NameError as e:

Variable name "e" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

            res["th_est1"] = None

        # estimate theta (from MLE)
        try:
            res["th_est2"] = popt[0] * sigma_est ** 2 / 2
        except NameError as e:

Variable name "e" doesn't conform to snake_case naming style ('([^\\W\\dA-Z][^\\WA-Z]2,|_[^\\WA-Z]*|__[^\\WA-Z\\d_][^\\WA-Z]+__)$' pattern)

            res["th_est2"] = None

        return res