You can use argrelextrema
for find local min and max:
from scipy.signal import argrelextrema
np.random.seed(0)
rs = np.random.randn(200)
xs = [0]
for r in rs:
xs.append(xs[-1] * 0.9 + r)
df = pd.DataFrame(xs, columns=['data'], index=pd.date_range('2000-01-01',periods=len(xs)))
n = 5 # number of points to be checked before and after
# Find local peaks
df['min'] = df.iloc[argrelextrema(df.data.values, np.less_equal,
order=n)[0]]['data']
df['max'] = df.iloc[argrelextrema(df.data.values, np.greater_equal,
order=n)[0]]['data']
df['min_date'] = df.index.where(df['min'].notna())
df['max_date'] = df.index.where(df['max'].notna())
print (df.head(15))
data min max min_date max_date
2000-01-01 0.000000 0.000000 NaN 2000-01-01 NaT
2000-01-02 1.764052 NaN NaN NaT NaT
2000-01-03 1.987804 NaN NaN NaT NaT
2000-01-04 2.767762 NaN NaN NaT NaT
2000-01-05 4.731879 NaN NaN NaT NaT
2000-01-06 6.126249 NaN 6.126249 NaT 2000-01-06
2000-01-07 4.536346 NaN NaN NaT NaT
2000-01-08 5.032800 NaN NaN NaT NaT
2000-01-09 4.378163 NaN NaN NaT NaT
2000-01-10 3.837128 NaN NaN NaT NaT
2000-01-11 3.864013 NaN NaN NaT NaT
2000-01-12 3.621656 3.621656 NaN 2000-01-12 NaT
2000-01-13 4.713764 NaN NaN NaT NaT
2000-01-14 5.003425 NaN NaN NaT NaT
2000-01-15 4.624757 NaN NaN NaT NaT
EDIT:
Solution from real data:
df['Date'] = pd.to_datetime(df['Date'])
df = df.set_index('Date')
from scipy.signal import argrelextrema
n = 5
s1 = df.iloc[argrelextrema(df.Mean.values, np.less_equal,
order=n)[0]]['Mean']
s2 = df.iloc[argrelextrema(df.Mean.values, np.greater_equal,
order=n)[0]]['Mean']
s = s1.append(s2).sort_index()
print (s)
Date
2016-05-18 0.293171
2016-11-04 0.692509
2017-05-13 0.232963
2017-09-10 0.675797
2017-11-09 0.528592
2018-04-03 0.189523
2018-11-09 0.713351
Name: Mean, dtype: float64
s.to_csv('out.csc')