CODEWITHSUNDEEP
pythonpandasnumpy
mike01010
mike01010

Reputation: 6048

Pandas Rolling window - find date or number of actual days for min/max window

i have a time series that look like this

                    open          high           low         close
date
1982-10-06   191.98   191.98   191.98   191.98    191.98
1982-10-07   195.59   195.59   195.59   195.59    195.59
1982-10-08   198.89   198.89   198.89   198.89    198.89
...

I find them minimum value over that time series for window = 20.  That low can happen on any of the previous 19 days.  id like to add to the dataframe either the date or number of days back that that low occurred. Can anyone help with this?

df['window_low'] =  df['low'].rolling(min_periods=1, window=20, center=False).min()
df['low_date'] = ?
df['number_of_days_back'] = ?

Upvotes: 0

Views: 666

Answers (1)

BENY
BENY

Reputation: 323266

You can do with idxmin

df = df.reset_index()
df['date'] = pd.to_datetime(df['date'])
id = df['low'].rolling(min_periods=1, window=20, center=False).apply(lambda x: x.idxmin())
df['low_date'] = df['date'].reindex(id).values
df['number_of_days_back'] = df['date'] - df['low_date']

Upvotes: 1

Related Questions