How do I find the length of a dataframe in dask?

Question

How do I find the length of a dataframe using dask?

For example in pandas, I can do:

import pandas as pd
import numpy as np
df = pd.DataFrame(np.random.normal(0, 1, (5, 2)), columns=["A", "B"])
print df['A'].count()
print df

Output:

5
          A         B
0  1.538531  0.424717
1 -0.929843  1.323648
2 -1.283680  0.056199
3 -0.641035 -1.998241
4 -0.058598 -1.400637

In dask I try:

import dask.dataframe as dd
df_dask = dd.from_pandas(df, npartitions=3)
print df_dask
print df_dask['A'].count()

Output:

                     A        B
npartitions=2                  
0              float64  float64
2                  ...      ...
4                  ...      ...
Dask Name: from_pandas, 2 tasks

dd.Scalar<series-..., dtype=int32>

The real reason I need length is because df_dask.sample() takes a fraction and I want to sample a specified number of entries from the dataframe. I use length to compute this fraction. Is there an easier/faster way of doing this?

Answer 1

You can use len for length of dask DataFrame column or index:

print (len(df_dask['A']))
5

print (len(df_dask.index))
5

Your solution is beter if need count all non NaNs values - add compute:

df = pd.DataFrame(np.random.normal(0, 1, (5, 2)), columns=["A", "B"])
df.loc[0, 'A'] = np.nan
print (df)
          A         B
0       NaN -1.727669
1 -0.390900  0.573806
2  0.338589 -0.011830
3  2.392365  0.412912
4  0.978736  2.238143

import dask.dataframe as dd
df_dask = dd.from_pandas(df, npartitions=3)

print (df_dask['A'].count().compute())
4