Finding the local maxima and local minima in the data python-CodePudding

Data:

 --------------------- ------------------ --  
|      date_add       |           fnv_wa |  |
 --------------------- ------------------ -- 
| 2022-06-24 06:00:16 | 46.216866        |  |
| 2022-06-24 07:00:16 | 46.216866        |  |
| 2022-06-24 08:00:16 | 45.685139        |  |
| 2022-06-24 09:00:16 | 45.633936        |  |
| 2022-06-24 10:00:16 | 43.487337        |  |
| 2022-06-24 11:00:16 | 40.182756        |  |
| 2022-06-24 12:00:16 | 40.017330        |  |
| 2022-06-24 13:00:16 | 39.548623        |  |
| 2022-06-24 14:00:16 | 39.548623        |  |
| 2022-06-24 15:00:16 | 38.607271        |  |
| 2022-06-24 16:00:16 | 39.989759        |  |
| 2022-06-24 17:00:16 | 39.111426        |  |
| 2022-06-24 18:00:16 | 37.862854        |  |
| 2022-06-24 19:00:16 | 37.862854        |  |
| 2022-06-24 20:00:16 | 37.862854        |  |
| 2022-06-24 21:00:16 | 36.173146        |  |
| 2022-06-24 22:00:16 | 35.164835        |  |
 --------------------- ------------------ --

I'm trying to find all the local maxima's and minima's in my data, the approaches I tried are listed below:

Approach 1: Using scipy.signal's argrelextremafrom to find the local maxima's and minima's, but the limitation is when the data window is large, its now able to identify.

Implementation:

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']

Approach 2: Using dataframes shift functionality:

Implementation:

df['min'] = df.data[(df.data.shift(1) > df.data) & (df.data.shift(-1) > df.data)]
df['max'] = df.data[(df.data.shift(1) < df.data) & (df.data.shift(-1) < df.data)]

The problem with above two approaches, the local maxima at X~10 is not detected.

Please suggest an approach that can find all the local maxima and local minima in my data.

CodePudding user response：

Which value to you use for n?

Your code is working quite fine with n=3:

from scipy.signal import argrelextrema
n = 3
df['min'] = df.iloc[argrelextrema(df['fnv_wa'].values, np.less_equal,
                    order=n)[0]]['fnv_wa']
df['max'] = df.iloc[argrelextrema(df['fnv_wa'].values, np.greater_equal,
                    order=n)[0]]['fnv_wa']

ax = df.plot(y='fnv_wa')
df.plot(y='max', marker='o', color='g', ax=ax)
df.plot(y='min', marker='o', color='r', ax=ax)

output:

With n=2:

CodePudding user response：

Rudimentary approach (will probably perform poorly at scale)

local_minima = []
local_maxima = []
for i, row in df.iterrows():
    if i > 0 and i < len(df)-1:
        if df.loc[i,"fnv_wa"] < df.loc[i-1,"fnv_wa"] and df.loc[i,"fnv_wa"] < df.loc[i 1,"fnv_wa"]:
            local_minima.append(i)
        elif df.loc[i,"fnv_wa"] > df.loc[i-1,"fnv_wa"] and df.loc[i,"fnv_wa"] > df.loc[i 1,"fnv_wa"]:
            local_maxima.append(i)