CODEWITHSUNDEEP
pythonpandas
Diego Wu
Diego Wu

Reputation: 23

python pandas method "ffill"

This is my initial df 

       China  USA
2009    NaN   -9
2010    1.2    8
2011    1.5  NaN
2014    5.0  NaN
2015    NaN    8

and then I used reindex method to add some index

frame2=frame1.reindex([2009,2010,2011,2012,2013,2014,2015,2016],method="ffill")

I expected frame2 to pass from

       China  USA
2009    NaN   -9
2010    1.2    8
2011    1.5  NaN
2012    NaN  NaN
2013    NaN  NaN
2014    5.0  NaN
2015    NaN    8
2016    NaN  NaN

to

       China  USA
2009    NaN   -9
2010    1.2    8
2011    1.5    8
2012    1.5    8
2013    1.5    8
2014    5.0    8
2015    5.0    8
2016    5.0    8

but it returned

       China  USA
2009    NaN   -9
2010    1.2    8
2011    1.5  NaN
2012    1.5  NaN
2013    1.5  NaN
2014    5.0  NaN
2015    NaN    8
2016    NaN    8

it only filled some entries of my df..

Upvotes: 2

Views: 71

Answers (1)

piRSquared
piRSquared

Reputation: 294338

In a sense, it did forward fill. It brought forward the NaN from 2015 to 2016 in China as well as from 2011 through 2013 in USA.

Instead, do a ffill after the reindex

frame1.reindex([2009,2010,2011,2012,2013,2014,2015,2016]).ffill()

      China  USA
2009    NaN -9.0
2010    1.2  8.0
2011    1.5  8.0
2012    1.5  8.0
2013    1.5  8.0
2014    5.0  8.0
2015    5.0  8.0
2016    5.0  8.0

Further explanation

# Original

       China  USA
2009    NaN   -9
2010    1.2    8
2011    1.5  NaN
2014    5.0  NaN
2015    NaN    8

Your method did what it was supposed to. If took values found at existing indices and propagated them forward for all new indices.

frame1.reindex([2009,2010,2011,2012,2013,2014,2015,2016],method="ffill")

                            China  USA
2009                          NaN -9.0
2010                          1.2  8.0
2011                          1.5  NaN  <- From Original
2012                          1.5  NaN  <- And forward filled
2013                          1.5  NaN  <- And here
2014                          5.0  NaN  <- was there in original
2015        From Original ->  NaN  8.0
2016   And forward filled ->  NaN  8.0

Upvotes: 3

Related Questions