How to split dataframe into multiple sub-dataframes based on column value

grp = df['Column2'].ne(df['Column2'].shift()).cumsum()
cond = df['Column2'].ne(0)
out = [d for _, d in df[cond].groupby(grp)]

out

[   Column1  Column2
 0       13        1
 1       12        1,

    Column1  Column2
 4       15        1
 5       14        1
 6       12        1,

    Column1  Column2
 8       21        1
 9       45        1]

Another possible solution:

m = df['Column2'].eq(1)
starts = df['Column2'].diff().fillna(1).eq(1) & m
df['grp_id'] = starts.cumsum()
[df[(df['grp_id'].eq(i)) & m].drop(columns='grp_id') 
 for i in df['grp_id'][m].unique()]

This solution identifies contiguous measurement blocks by first creating a boolean mask of rows where Column2.eq(1), then detects the start of each block using diff() to find transitions from 0 to 1 (with fillna(1) handling the first row) combined with the mask to isolate only true beginnings; cumsum() converts these boolean start flags into sequential group IDs, and a list comprehension iterates over unique() group values to slice the dataframe via boolean indexing, producing separate sub-dataframes for each measurement block while dropping the temporary group column.

Output:

[   Column1  Column2
 0       13        1
 1       12        1,
    Column1  Column2
 4       15        1
 5       14        1
 6       12        1,
    Column1  Column2
 8       21        1
 9       45        1]

Here’s one way to solve it using pandas. The key idea is to assign a “measurement number” to each consecutive block of 1s in Column2. Then, if you want, you can split them into separate DataFrames.

import pandas as pd

# Sample DataFrame
data = {
    "Column1": [13,12,15,16,15,14,12,11,21,45,44],
    "Column2": [1,1,0,0,1,1,1,0,1,1,0]
}

df = pd.DataFrame(data)

# Step 1: Add a new column for measurement numbers
df['Measurement'] = 0

measurement = 0       # Counter for each measurement
in_measurement = False  # Flag to track if we are inside a block of 1s

# Step 2: Iterate over rows to assign measurement numbers
for i in range(len(df)):
    if df.loc[i, 'Column2'] == 1:
        if not in_measurement:
            measurement += 1   # Start of a new measurement
            in_measurement = True
        df.loc[i, 'Measurement'] = measurement
    else:
        in_measurement = False  # End of the current measurement

print(df)

Output:

    Column1  Column2  Measurement
0        13        1            1
1        12        1            1
2        15        0            0
3        16        0            0
4        15        1            2
5        14        1            2
6        12        1            2
7        11        0            0
8        21        1            3
9        45        1            3
10       44        0            0

Split into separate DataFrames

Once you have the Measurement column, you can split the DataFrame like this:

# Get unique measurement numbers
measurements = df['Measurement'].unique()

# Create a list of sub-DataFrames
dfs = [df[df['Measurement']==m].drop(columns='Measurement') for m in measurements if m != 0]

# Example: first measurement
df2 = dfs[0]
print(df2)

This will give you:

   Column1  Column2
0       13        1
1       12        1

You can similarly access df3, df4, etc.

How this works

1. in_measurement keeps track of whether you are currently inside a block of 1s.

2. measurement increments only when a new block starts.

3. Every 1 in the same block gets the same number.

4. 0s are ignored but mark the end of a block.

5. After that, splitting into sub-DataFrames is easy.