Fuzzy String match and merge database - Dataframe - python

I have two dataframes (with strings) that I am trying to compare to each other. One has a list of areas, the other has a list of areas with long,lat info as well. I am struggling to write a code to perform the following:
1) Check if the string in df1 matches (or a partially matches) area names in df2, then it will merge & carry over the long lat columns.
2) if df1 does not match with df2, then the new column will have NaN or zero.
Code:
import pandas as pd
df1 = pd.read_csv('Dubai Communities1.csv')
df1.head()
CNAME_E1
0 Abu Hail
1 Al Asbaq
2 Al Aweer First
3 Al Aweer Second
4 Al Bada
df2 = pd.read_csv('Dubai Communities2.csv')
df2.head()
COMM_NUM CNAME_E2 Latitude Longitude
0 315 UMM HURAIR 55.3237 25.2364
1 917 AL MARMOOM 55.4518 24.9756
2 624 WARSAN 55.4034 25.1424
3 123 AL MUTEENA 55.3228 25.2739
4 813 AL ROWAIYAH 55.3981 25.1053
The output after search and join will look like this:
CName_E1 CName_E3 Latitude Longitude
0 Area1 Area1 22 7.25
1 Area2 Area2 38 71.83
2 Area3 NaN NaN NaN
3 Area4 Area4 35 8.05

Related

How to spread the data in pandas?

i'm working on spread r equivalent in pandas my dataframe looks like below
Name age Language year Period
Nik 18 English 2018 Beginer
John 19 French 2019 Intermediate
Kane 33 Russian 2017 Advanced
xi 44 Thai 2015 Beginer
and looking for output like this
Name age Language Beginer Intermediate Advanced
Nik 18 English 2018
John 19 French 2019
Kane 33 Russian 2017
John 44 Thai 2015
my code
pd.pivot(x1,values='year', columns=['Period'])
i'm getting only these columns Beginer,Intermediate,Advanced not the entire dataframe
while reshaping it i tried using index but says no duplicates in index.
So i created new index column but still not getting entire dataframe
If I understood correctly you could do something like this:
# create dummy columns
res = pd.get_dummies(df['Period']).astype(np.int64)
res.values[np.arange(len(res)), np.argmax(res.values, axis=1)] = df['year']
# concat and drop columns
output = pd.concat((df.drop(['year', 'Period'], 1), res), 1)
print(output)
Output
Name age Language Advanced Beginner Intermediate
0 Nik 18 English 0 2018 0
1 John 19 French 0 0 2019
2 Kane 33 Russian 2017 0 0
3 xi 44 Thai 0 2015 0
If you want to match the exact same output, convert the column to categorical first, and specify the order:
# encode as categorical
df['Period'] = pd.Categorical(df['Period'], ['Beginner', 'Advanced', 'Intermediate'], ordered=True)
# create dummy columns
res = pd.get_dummies(df['Period']).astype(np.int64)
res.values[np.arange(len(res)), np.argmax(res.values, axis=1)] = df['year']
# concat and drop columns
output = pd.concat((df.drop(['year', 'Period'], 1), res), 1)
print(output)
Output
Name age Language Beginner Advanced Intermediate
0 Nik 18 English 2018 0 0
1 John 19 French 0 0 2019
2 Kane 33 Russian 0 2017 0
3 xi 44 Thai 2015 0 0
Finally if you want to replace the 0, with missing values, add a third step:
# create dummy columns
res = pd.get_dummies(df['Period']).astype(np.int64)
res.values[np.arange(len(res)), np.argmax(res.values, axis=1)] = df['year']
res = res.replace(0, np.nan)
Output (with missing values)
Name age Language Beginner Advanced Intermediate
0 Nik 18 English 2018.0 NaN NaN
1 John 19 French NaN NaN 2019.0
2 Kane 33 Russian NaN 2017.0 NaN
3 xi 44 Thai 2015.0 NaN NaN
One way you can get to the equivalent of R's spread function using pd.pivot_table:
If you don't mind about the index, you can use reset_index() on the newly created df:
new_df = (pd.pivot_table(df, index=['Name','age','Language'],columns='Period',values='year',aggfunc='sum')).reset_index()
which will get you:
Period Name age Language Advanced Beginer Intermediate
0 John 19 French NaN NaN 2019.0
1 Kane 33 Russian 2017.0 NaN NaN
2 Nik 18 English NaN 2018.0 NaN
3 xi 44 Thai NaN 2015.0 NaN
EDIT
If you have many columns in your dataframe and you want to include them in the reshaped dataset:
Grab in a list the columns to be used in pivot table (i.e. Period and year)
Grab all the other columns in your dataframe in a list (using not in)
Use the index_cols as index in the pd.pivot_table() command
non_index_cols = ['Period','year'] # SPECIFY THE 2 COLUMNS IN THE PIVOT TABLE TO BE USED
index_cols = [i for i in df.columns if i not in non_index_cols] # GET ALL THE REST IN A LIST
new_df = (pd.pivot_table(df, index=index_cols,columns='Period',values='year',aggfunc='sum')).reset_index()
The new_df, will include all the columns of your initial dataframe.

Python Pandas fill missing zipcode with values from another datafrane based on conditions

I have a dataset in which I add coordinates to cities based on zip-codes but several of these zip-codes are missing. Also, in some cases cities are missing, states are missing, or both are missing. For example:
ca_df[['OWNER_CITY', 'OWNER_STATE', 'OWNER_ZIP']]
OWNER_CITY OWNER_STATE OWNER_ZIP
495 MIAMI SHORE PA
496 SEATTLE
However, a second dataset has city, state & the matching zip-codes. This one is complete without any missing values.
df_coord.head()
OWNER_ZIP CITY STATE
0 71937 Cove AR
1 72044 Edgemont AR
2 56171 Sherburn MN
I want to fill in the missing zip-codes in the first dataframe if:
Zip-code is empty
City is present
State is present
This is an all-or-nothing operations means, either all three criteria are met and the zip-code gets filled or nothing changes.
However, this is a fairly large dataset with > 50 million records so ideally I want to vectorize the operation by working column-wise.
Technically, that would fit np.where but as far as I know, np.where only takes of condition in the following format:
df1['OWNER_ZIP'] = np.where(df["cond"] ==X, df_coord['OWNER_ZIP'], "")
How do I ensure I only fill missing zip-codes when all conditions are met?
Given ca_df:
OWNER_CITY OWNER_STATE OWNER_ZIP
0 Miami Shore Florida 111
1 Los Angeles California NaN
2 Houston NaN NaN
and df_coord:
OWNER_ZIP CITY STATE
0 111 Miami Shore Florida
1 222 Los Angeles California
2 333 Houston Texas
You can use pd.notna along with pd.DataFrame#index like this:
inferrable_zips_df = pd.notna(ca_df["OWNER_CITY"]) & pd.notna(ca_df["OWNER_STATE"])
is_inferrable_zip = ca_df.index.isin(df_coord[inferrable_zips_df].index)
ca_df.loc[is_inferrable_zip, "OWNER_ZIP"] = df_coord["OWNER_ZIP"]
with ca_df resulting as:
OWNER_CITY OWNER_STATE OWNER_ZIP
0 Miami Shore Florida 111
1 Los Angeles California 222
2 Houston NaN NaN
I've changed the "" to np.nan, but if you still wish to use "" then you just need to change pd.notna(ca_df[...]) to ca_df[...] == "".
You can combine numpy.where statements to combine multiple rules. This should give you the array of row indices which abide to each of the three rules:
np.where(df["OWNER_ZIP"] == X) and np.where(df["CITY"] == Y) and np.where(df["STATE"] == Z)
Use:
print (df_coord)
OWNER_ZIP CITY STATE
0 71937 Cove AR
1 72044 Edgemont AR
2 56171 Sherburn MN
3 123 MIAMI SHORE PA
4 789 SEATTLE AA
print (ca_df)
OWNER_ZIP OWNER_CITY OWNER_STATE
0 NaN NaN NaN
1 72044 Edgemont AR
2 56171 NaN MN
3 NaN MIAMI SHORE PA
4 NaN SEATTLE NaN
First is necessary test if same dtypes in columns matching:
#or convert ca_df['OWNER_ZIP'] to integers
df_coord['OWNER_ZIP'] = df_coord['OWNER_ZIP'].astype(str)
print (df_coord.dtypes)
OWNER_ZIP object
CITY object
STATE object
dtype: object
print (ca_df.dtypes)
OWNER_ZIP object
OWNER_CITY object
OWNER_STATE object
dtype: object
Then filter for each combinations of columns - missing and non missing values and add new data by merge, then convert index to same like filtered data and assign back:
mask1 = ca_df['OWNER_CITY'].notna() & ca_df['OWNER_STATE'].notna() & ca_df['OWNER_ZIP'].isna()
df1 = ca_df[mask1].drop('OWNER_ZIP', axis=1).merge(df_coord.rename(columns={'CITY':'OWNER_CITY','STATE':'OWNER_STATE'})).set_index(ca_df.index[mask1])
ca_df.loc[mask1, ['OWNER_ZIP','OWNER_CITY','OWNER_STATE']] = df1
mask2 = ca_df['OWNER_CITY'].notna() & ca_df['OWNER_STATE'].isna() & ca_df['OWNER_ZIP'].isna()
df2 = ca_df[mask2].drop(['OWNER_ZIP','OWNER_STATE'], axis=1).merge(df_coord.rename(columns={'CITY':'OWNER_CITY','STATE':'OWNER_STATE'})).set_index(ca_df.index[mask2])
ca_df.loc[mask2, ['OWNER_ZIP','OWNER_CITY','OWNER_STATE']] = df2
mask3 = ca_df['OWNER_CITY'].isna() & ca_df['OWNER_STATE'].notna() & ca_df['OWNER_ZIP'].notna()
df3 = ca_df[mask3].drop(['OWNER_CITY'], axis=1).merge(df_coord.rename(columns={'CITY':'OWNER_CITY','STATE':'OWNER_STATE'})).set_index(ca_df.index[mask3])
ca_df.loc[mask3, ['OWNER_ZIP','OWNER_CITY','OWNER_STATE']] = df3
print (ca_df)
OWNER_ZIP OWNER_CITY OWNER_STATE
0 NaN NaN NaN
1 72044 Edgemont AR
2 56171 Sherburn MN
3 123 MIAMI SHORE PA
4 789 SEATTLE AA
You can do a left join on these dataframes considering join on the columns 'city' and 'state'. That would give you the zip-code corresponding to a city and state if both values are non-null in the first dataframe (OWNER_CITY, OWNER_STATE, OWNER_ZIP) and since it would be a left join, it would also preserve your rows which either don't have a zip-code or have null/empty city and state values.

Adding a column with values from another dataframe based on column conditions

I have two dataframes of differing index lengths that look like this:
df_1:
State Month Total Time ... N columns
AL 4 1000
AL 5 500
.
.
.
VA 11 750
VA 12 1500
df_2:
State Month ... N columns
AL 4
AL 5
.
.
.
VA 11
VA 12
I would like to add a Total Time column to df_2 that uses the values from the Total Time column of df_1 if the State and Month value are the same between dataframes. Ultimately, I would end up with:
df_2:
State Month Total Time ... N columns
AL 4 1000
AL 5 500
.
.
.
VA 11 750
VA 12 1500
I want to do this conditionally since the index lengths are not the same. I have tried this so far:
def f(row):
if (row['State'] == row['State']) & (row['Month'] == row['Month']):
val = x for x in df_1["Total Time"]
return val
df_2['Total Time'] = df_2.apply(f, axis=1)
This did not work. What method would you use to accomplish this?
Any help is appreciated!
You can do this:
Consider my sample dataframes:
In [2327]: df_1
Out[2327]:
State Month Total Time
0 AL 2 1000
1 AB 4 500
2 BC 1 600
In [2328]: df_2
Out[2328]:
State Month
0 AL 2
1 AB 5
In [2329]: df_2 = pd.merge(df_2, df_1, on=['State', 'Month'], how='left')
In [2330]: df_2
Out[2330]:
State Month Total Time
0 AL 2 1000.0
1 AB 5 NaN
As mentioned in other comment, pd.merge() is how you would join two dataframes and extract a column. The issue is that merging solely on 'State' and 'Month' would result in every permutation becoming a new column (all Al-4 in df_1 would join with all other AL-4 in df_2).
With your example, there'd be
df_1
State Month Total Time df_1 col...
0 AL 4 1000 6
1 AL 4 500 7
2 VA 12 750 8
3 VA 12 1500 9
df_2
State Month df_2 col...
0 AL 4 1
1 AL 4 2
2 VA 12 3
3 VA 12 4
df_1_cols_to_use = ['State', 'Month', 'Total Time']
# note the selection of the column to use from df_1. We only want the column
# we're merging on, plus the column(s) we want to bring in, in this case 'Total Time'
new_df = pd.merge(df_2, df_1[df_1_cols_to_use], on=['State', 'Month'], how='left')
new_df:
State Month df_2 col... Total Time
0 AL 4 1 1000
1 AL 4 1 500
2 AL 4 2 1000
3 AL 4 2 500
4 VA 12 3 750
5 VA 12 3 1500
6 VA 12 4 750
7 VA 12 4 1500
You mention these have differing index lengths. Based on the parameters of the question, it's not possible to determine what value of Total Time would match up with a row in df_2. If there's three AL-4 entries in df_2, do they each get 1000, 500, or some combination? That info would be needed. Without this, this would be the best guess at getting all possibilities.

Multiply columns based on two columns conditions from different dataframes?

I have two dataframes as indicated below:
dfA =
Country City Pop
US Washington 1000
US Texas 5000
CH Geneva 500
CH Zurich 500
dfB =
Country City Density (pop/km2)
US Washington 10
US Texas 50
CH Geneva 5
CH Zurich 5
What I want is to compare the columns Country and City from both dataframes, and when these match such as:
US Washington & US Washington in both dataframes, it takes the Pop value and divides it by Density, as to get a new column area in dfB with the resulting division. Example of first row results dfB['area km2'] = 100
I have tried with np.where() but it is nit working. Any hints on how to achieve this?
Using index matching and div
match_on = ['Country', 'City']
dfA = dfA.set_index(match_on)
dfA.assign(ratio=dfA.Pop.div(df.set_index(['Country', 'City'])['Density (pop/km2)']))
Country City
US Washington 100.0
Texas 100.0
CH Geneva 100.0
Zurich 100.0
dtype: float64
You can also use merge to combine the two dataframes and divide as usual:
dfMerge = dfA.merge(dfB, on=['Country', 'City'])
dfMerge['area'] = dfMerge['Pop'].div(dfMerge['Density (pop/km2)'])
print(dfMerge)
Output:
Country City Pop Density (pop/km2) area
0 US Washington 1000 10 100.0
1 US Texas 5000 50 100.0
2 CH Geneva 500 5 100.0
3 CH Zurich 500 5 100.0
you can also use merge like below
dfB["Area"] = dfB.merge(dfA, on=["Country", "City"], how="left")["Pop"] / dfB["Density (pop/km2)"]
dfB

How to extract info from original dataframe after doing some analysis on it?

So I had a dataframe and I had to do some cleansing to minimize the duplicates. In order to do that I created a dataframe that had instead of 40 only 8 of the original columns. Now I have two columns I need for further analysis from the original dataframe but they would mess with the desired outcome if I used them in my previous analysis. Anyone have any idea on how to "extract" these columns based on the new "clean" dataframe I have?
You can merge the new "clean" dataframe with the other two variables by using the indexes. Let me use a pratical example. Suppose the "initial" dataframe, called "df", is:
df
name year reports location
0 Jason 2012 4 Cochice
1 Molly 2012 24 Pima
2 Tina 2013 31 Santa Cruz
3 Jake 2014 2 Maricopa
4 Amy 2014 3 Yuma
while the "clean" dataframe is:
d1
year location
0 2012 Cochice
2 2013 Santa Cruz
3 2014 Maricopa
The remaing columns are saved in dataframe "d2" ( d2 = df[['name','reports']] ):
d2
name reports
0 Jason 4
1 Molly 24
2 Tina 31
3 Jake 2
4 Amy 3
By using the inner join on the indexes d1.merge(d2, how = 'inner' left_index= True, right_index = True) you get the following result:
name year reports location
0 Jason 2012 4 Cochice
2 Tina 2013 31 Santa Cruz
3 Jake 2014 2 Maricopa
You can make a new dataframe with the specified columns;
import pandas
#If your columns are named a,b,c,d etc
df1 = df[['a','b']]
#This will extract columns 0, to 2 based on their index
#[remember that pandas indexes columns from zero!
df2 = df.iloc[:,0:2]
If you could, provide a sample piece of data, that'd make it easier for us to help you.

Categories

Resources