n1 = DataFrame({'zhanghui':[1,2,3,4] , 'wudi':[17,'gx',356,23] ,'sas'[234,51,354,123] })
n2 = DataFrame({'zhanghui_x':[1,2,3,5] , 'wudi':[17,23,'sd',23] ,'wudi_x':[17,23,'x356',23] ,'wudi_y':[17,23,'y356',23] ,'ddd':[234,51,354,123] })
code above defined two DataFrame objects. I wanna use 'zhanghui' field from n1 and 'zhanghui_x' field from n2 as "on" field merge n1 and n2,so my code like this:
n1.merge(n2,how = 'inner',left_on = 'zhanghui',right_on='zhanghui_x')
and then result columns given like this :
sas wudi_x zhanghui ddd wudi_y wudi_x wudi_y zhanghui_x
Some duplicate columns appeared,such as 'wudi_x' ,'wudi_y'.
So it's a pandas inner problems or I had a wrong usage about pd.merge ?
From pandas documentation, the merge() function has following properties;
pd.merge(left, right, how='inner', on=None, left_on=None, right_on=None,
left_index=False, right_index=False, sort=True,
suffixes=('_x', '_y'), copy=True, indicator=False,
validate=None)
where suffixes denote default suffix string to be attached to 'over-lapping' columns with defaults '_x' and '_y'.
I'm not sure if I understood your follow-up question correctly, but;
#case1
if the first dataFrame has column 'column_name_x' and the second dataFrame has column 'column_name' then there are no over-lapping columns and therefore no suffixes are attached.
#case2
if the first dataFrame has columns 'column_name', 'column_name_x' and the second dataFrame also has column 'column_name', the default suffixes attach to over-lapping columns and therefore the first frame's 'columnn_name' becomes 'column_name_x' and result in a duplicate of already existing column.
You can however, pass a None value to one(not all) of the suffixes to ensure that column names of certain dataFrame remain as-is.
Your approach is right, pandas automatically gives postscripts after merging the columns that are "duplicated" with the original headers given a postscript _x, _y, etc.
you can first select what columns to merge and proceed:
cols_to_use = n2.columns - n1.columns
n1.merge(n2[cols_to_use],how = 'inner',left_on = 'zhanghui',right_on='zhanghui_x')
result columns:
sas wudi zhanghui ddd wudi_x wudi_y zhanghui_x
When I tried to run cols_to_use = n2.columns - n1.columns,it gave me a TypeError like this:
cannot perform __sub__ with this index type: <class pandas.core.indexes.base.Index'>
then I tried to use code below:
cols_to_use = [i for i in list(n2.columns) if i not in list(n1.columns) ]
It worked fine,result columns given like this:
sas wudi zhanghui ddd wudi_x wudi_y zhanghui_x
So,#S Ringne's method really resolved my problems.
=============================================
Pandas just simply add suffix such as '_x' to resolve the duplicate-column-name problem when it comes to merging two Frame objects.
But what will it happen if the name form of 'a-column-name'+'_x' appears in either Frame object? I used to think that it will check if the name form of 'a-column-name'+'_x' appears, But actually pandas doesn't have this check?
Related
I am extracting tables from pdf using Camelot. Two of the columns are getting merged together with a newline separator. Is there a way to separate them into two columns?
Suppose the column looks like this.
A\nB
1\n2
2\n3
3\n4
Desired output:
|A|B|
|-|-|
|1|2|
|2|3|
|3|4|
I have tried df['A\nB'].str.split('\n', 2, expand=True) and that splits it into two columns however I want the new column names to be A and B and not 0 and 1. Also I need to pass a generalized column label instead of actual column name since I need to implement this for several docs which may have different column names. I can determine such column name in my dataframe using
colNew = df.columns[df.columns.str.contains(pat = '\n')]
However when I pass colNew in split function, it throws an attribute error
df[colNew].str.split('\n', 2, expand=True)
AttributeError: DataFrame object has no attribute 'str'
You can take advantage of the Pandas split function.
import pandas as pd
# recreate your pandas series above.
df = pd.DataFrame({'A\nB':['1\n2','2\n3','3\n4']})
# first: Turn the col into str.
# second. split the col based on seperator \n
# third: make sure expand as True since you want the after split col become two new col
test = df['A\nB'].astype('str').str.split('\n',expand=True)
# some rename
test.columns = ['A','B']
I hope this is helpful.
I reproduced the error from my side... I guess the issue is that "df[colNew]" is still a dataframe as it contains the indexes.
But .str.split() only works on Series. So taking as example your code, I would convert the dataframe to series using iloc[:,0].
Then another line to split the column headers:
df2=df[colNew].iloc[:,0].str.split('\n', 2, expand=True)
df2.columns = 'A\nB'.split('\n')
I have a very large data frame that I want to split ALL of the columns except first two based on a comma delimiter. So I need to logically reference column names in a loop or some other way to split all the columns in one swoop.
In my testing of the split method:
I have been able to explicitly refer to ( i.e. HARD CODE) a single column name (rs145629793) as one of the required parameters and the result was 2 new columns as I wanted.
See python code below
HARDCODED COLUMN NAME --
df[['rs1','rs2']] = df.rs145629793.str.split(",", expand = True)
The problem:
It is not feasible to refer to the actual column names and repeat code.
I then replaced the actual column name rs145629793 with columns[2] in the split method parameter list.
It results in an ERROR
'str has ni str attribute'
You can index columns by position rather than name using iloc. For example, to get the third column:
df.iloc[:, 2]
Thus you can easily loop over the columns you need.
I know what you are asking, but it's still helpful to provide some input data and expected output data. I have included random input data in my code below, so you can just copy and paste this to run, and try to apply it to your dataframe:
import pandas as pd
your_dataframe=pd.DataFrame({'a':['1,2,3', '9,8,7'],
'b':['4,5,6', '6,5,4'],
'c':['7,8,9', '3,2,1']})
import copy
def split_cols(df):
dict_of_df = {}
cols=df.columns.to_list()
for col in cols:
key_name = 'df'+str(col)
dict_of_df[key_name] = copy.deepcopy(df)
var=df[col].str.split(',', expand=True).add_prefix(col)
df=pd.merge(df, var, how='left', left_index=True, right_index=True).drop(col, axis=1)
return df
split_cols(your_dataframe)
Essentially, in this solution you create a list of the columns that you want to loop through. Then you loop through that list and create new dataframes for each column where you run the split() function. Then you merge everything back together on the index. I also:
included a prefix of the column name, so the column names did not have duplicate names and could be more easily identifiable
dropped the old column that we did the split on.
Just import copy and use the split_cols() function that I have created and pass the name of your dataframe.
Given a pandas DataFrame (df), where one column (unique_val_col) should have a unique value, what is the best the best way to extract this value (not as a list)?
So far I've used the following code:
output = list(set(df[unique_val_col)))
if len(output)==1: output = output[0]
Or if there is a chance for nans then change the first line to be:
output = [val for val in list(set(df[unique_val_col))) if val == val]
The question is whether there is a more direct way, that would also reflect the fact that the column actually has only one value without needing the 'if' statement.
I think you are trying to find a value that occurs only once, if that's so you could achieve it like this
df['unique_value_counts'].value_counts().sort_values(ascending=False).keys()[0]
I have a column of datetimes and need to change several of these values to new datetimes. When I set the values using df.loc[indices, 'col'] = new_datetimes, the unaffected values are coerced to int while the new set values are in datetime. If I set the values one at a time, no type coercion occurs.
For illustration I created a sample df with just one column.
df = pd.DataFrame([dt.datetime(2019,1,1)]*5)
df.loc[[1,3,4]] = [dt.datetime(2019,1,2)]*3
df
This produces the following:
output
If I change indices 1,3,4 individually:
df = pd.DataFrame([dt.datetime(2019,1,1)]*5)
df.loc[1] = dt.datetime(2019,1,2)
df.loc[3] = dt.datetime(2019,1,2)
df.loc[4] = dt.datetime(2019,1,2)
df
I get the correct output:
output
A suggestion was to turn the list to a numpy array before setting, which does resolve the issue. However, if you try to set multiple columns (some of which are not datetime) using a numpy array, The issue arises again.
In this example the dataframe has two columns and I try to set both columns.
df = pd.DataFrame({'dt':[dt.datetime(2019,1,1)]*5, 'value':[1,1,1,1,1]})
df.loc[[1,3,4]] = np.array([[dt.datetime(2019,1,2)]*3, [2,2,2]]).T
df
This gives the following output:
output
Can someone please explain what is causing the coercion and how to prevent it from doing so? The code I wrote that uses this was written over a month ago and used to work just fine, could it be one of those warnings about future version of pandas deprecating certain functionalities?
An explanation of what is going on would be greatly appreciated because I wrote a other codes that likely employ similar functionality want to make sure everything works as intended.
The solution proposed by w-m has such an "awkward detail" than
the result column has also the time part (it didn't have it
before).
I have also such a remark, that DataFrames are tables not Series,
so they have columns, each with its name and it is a bad habit to
rely on default column names (consecutive numbers).
So I propose another solution, addressing both above issues:
To create the source DataFrame I executed:
df = pd.DataFrame([dt.datetime(2019, 1, 1)]*5, columns=['c1'])
Note that I provided a name for the only column.
Then I created another DataFrame:
df2 = pd.DataFrame([dt.datetime(2019,1,2)]*3, columns=['c1'], index=[1,3,4])
It contains your "new" dates and the numbers which you used in loc
I set as the index (again with the same column name).
Then, to update df, use (not surprisingly) df.update:
df.update(df2)
This function performs in-place update, so if you print(df), you will get:
c1
0 2019-01-01
1 2019-01-02
2 2019-01-01
3 2019-01-02
4 2019-01-02
As you can see, under indices 1, 3 and 4 you have new dates
and there is no time part, just like before.
[dt.datetime(2019,1,2)]*3 is a Python list of objects. This particular list happens to contain only datetimes, but Pandas does not seem to recognize that, and treats it as it is - a list of any kind of objects.
If you convert it into a typed array, then Pandas will keep the original dtype of the column intact:
df.loc[[1,3,4]] = np.asarray([dt.datetime(2019,1,2)]*3)
I hope this workaround helps you, but you may still want to file a bug with Pandas. I don't have an explanation as to why the datetime objects should be coerced to ints in the first output example.
I am working with pandas / python /numpy / datalab/bigQuery to generate an input table for machine learning processing. The data is genomic - and right now, I am working with small subset of
174 rows
12430 columns
The column names are extracted from bigQuery (df_pik3ca_features = bq.Query(std_sql_features).to_dataframe(dialect='standard',use_cache=True))
at the same way, the row names are extracted: samples_rows = bq.Query('SELECT sample_id FROMspeedy-emissary-167213.pgp_orielresearch.pgp_PIK3CA_all_features_values_step_3GROUP BY sample_id')
what would be the easiest way to create a dataframe / matrix with named rows and columns that were extracted.
I explored the dataframes in pandas and could not find the way to pass the names as parameter.
for empty array, I was able to find the following (numpy) with no names:
a = np.full([num_of_rows, num_of_columns], np.nan)
a.columns
I know R very well (if there is no other way - I hope that I can use it with datalab)
any idea?
Many thanks!
If you have your column names and row names stored in lists then you can just use .loc to select the exact rows and columns you desire. Just make sure that the row names are in the index. You might need to do df.set_index('sample_id') to put the correct row name in the index.
Assuming the rows and columns are in variables row_names and col_names, do this.
df.loc[row_names, col_names]