[python]常用总结

splitting numpy array based on value

https://stackoverflow.com/questions/38277182/splitting-numpy-array-based-on-value

import numpy as np
from itertools import groupby
b = np.array([  1,   1, 100,   1,   1,   3,   1,   1,   1])
c= [list(g) for k, g in groupby(b, lambda x: x != 100) if k]
#[[1, 1], [1, 1], [1], [1]]
d = [np.sum(k) for k in c]
# [2, 2, 1, 1]

[ [k,list(g)] for k, g in groupby(b, lambda x: x != 100)]
#[[True, [1, 1]], [False, [100]], [True, [1, 1]], [False, [100]], [True, [1]], [False, [100]], [True, [1]]]

http://blog.konghy.cn/2017/04/25/python-itertools/

groupby(iterable[, keyfunc])
对 iterable 中的元素进行分组。keyfunc 是分组函数，用于对 iterable 的连续项进行分组，如果不指定，则默认对 iterable 中的连续相同项进行分组，返回一个 (key, sub-iterator) 的迭代器。

csv

numpy.savetxt("foo.csv", a, delimiter=",")
numpy.savetxt("foo.csv", a, fmt="%s,%f,%d"delimiter=",")

存图

	fig = plt.figure(figsize=(16, 8))
    plt.plot(df["population"],'b',linewidth=3)
    plt.plot(df_pred["population"],'r')
    fig.savefig("img_order{}_ifres{}/{}_dayflag{}_num{}_predict".format(order_opt,ifres,mesh,dayflag,num_day), dpi=200, bbox_inches = 'tight')
    #plt.show()
    plt.close(fig)

    decomposition.plot()
    fig = plt.gcf()
    fig.savefig("img_order{}_ifres{}/{}_dayflag{}_num{}_decomposition{}".format(order_opt,ifres,mesh,dayflag,num_day,order_local), dpi=200, bbox_inches = 'tight')
    #plt.show()

颜色

有时候需要很多很多颜色又不想一个个自己定义，

from matplotlib import colors
colors_df = pd.DataFrame.from_dict(colors.cnames,orient='index')
colors_df.values

x轴每隔4个显示一次

plt.xticks(x1[::4],names[::4],rotation=90, fontsize=fs)