#!/usr/bin/env python
# coding: utf-8

# # Metrics widget
# 
# Execute this notebook to test out the different metrics. 

# In[ ]:


import numpy as np
import matplotlib.pyplot as plt
from ipywidgets import interact   # pip install ipywidgets
import modelskill.metrics as mtr


# In[ ]:


metrics = [mtr.bias, mtr.max_error, mtr.rmse, mtr.urmse, mtr.mae, mtr.mape, 
        mtr.mef, mtr.si, mtr.cc, mtr.spearmanr, mtr.r2, mtr.nse, mtr.willmott, mtr.lin_slope, mtr.kge, mtr.ev]


# In[ ]:


n = 50
x = np.linspace(0.0, 6.0, num=n)
y_obs = 2.0+3*np.sin(x/2.4)
noise_vec = np.random.randn(n)
noise_vec = noise_vec - noise_vec.mean()


# In[ ]:


def plot_metrics(bias, noise_level, fixed_y_axis=True):
    y_mod = y_obs + bias + noise_level*noise_vec
    plt.plot(x, y_obs, 'r.-', label="obs")
    plt.plot(x, y_mod, 'o-', label="model")
    plt.title(f"y_model = y_obs + {bias} + {noise_level}*noise")

    ymax = max(max(y_obs),max(y_mod))
    ymin = min(min(y_obs), min(y_mod))
    if fixed_y_axis:
        ymax = 8
        ymin = 1
    ystep = 1.2*(ymax - ymin)/len(metrics)
    ypos = ymax + 0.5
    for m in metrics:
        plt.text(6.5, ypos, f"{m.__name__}:")
        plt.text(8.0, ypos, f"{m(y_obs,y_mod):.4f}")
        ypos = ypos - ystep
    plt.legend(loc=2)
    if fixed_y_axis:
        plt.ylim(ymin, ymax)

    plt.show()


# In[ ]:


interact(plot_metrics, bias = (-1,3,0.1), noise_level=(0,2,0.05));