CODEWITHSUNDEEP
pythonpython-2.7pandasnumpy
Cyrille MODIANO
Cyrille MODIANO

Reputation: 2376

Finding a declining trend in a pandas dataframe

I have a dataframe with the following columns:

['NAME','SNAPDATE','RANK']

The dataframe can contains several occurrences of the same NAME at different dates with each time a different rank. I would like to find the product which rank improved (meaning the value declined)

I tried this:

diffs = df.RANK.diff()
df = df.assign(
     change=np.where(
     diffs > 0, 'increase', np.where(
     diffs < 0, 'decline', '------')))

The problem it is treating the dataframe as a whole and I need to find the trend for a single NAME.

Any idea.

Thanks

Upvotes: 0

Views: 338

Answers (1)

Konrad
Konrad

Reputation: 562

I think you can use sort_values together with groupby method:

df.sort_values(by=['NAME', 'SNAPDATE']).groupby(by='NAME').RANK.diff()

E.g.

df = pd.DataFrame({
    'NAME': ['Name1', 'Name1', 'Name2', 'Name2', 'Name1', 'Name2'],
    'SNAPDATE': ['2018-04-01', '2018-04-02', '2018-04-03', '2018-04-01', '2018-04-03', '2018-04-02'],
    'RANK': [10, 11, 12, 10, 9 ,8]
})

df.sort_values(by=['NAME', 'SNAPDATE']).groupby(by='NAME').RANK.diff()

Output:

0    NaN
1    1.0
2    1.0
3    NaN
4   -1.0
5   -1.0
Name: RANK, dtype: float64

Upvotes: 1

Related Questions