Datasets:

himanshu-bhatnagar-exl-2302

/

SAS_to_Python_clean_v1

like 1

PROC IMPORT DATAFILE = "/path/to/dataset.csv" DBMS = CSV OUT = WORK.MYDATA; RUN;

import pandas as pd df = pd.read_csv('/path/to/dataset.csv')

PROC EXPORT DATA = WORK.MYDATA OUTFILE = "/path/to/output.csv" DBMS = CSV; RUN;

df.to_csv('/path/to/output.csv')

PROC PRINT DATA = WORK.MYDATA(OBS=5); RUN;

df.head(5)

PROC CONTENTS DATA = WORK.MYDATA; RUN;

df.info()

PROC SQL; SELECT COUNT(*) FROM WORK.MYDATA; QUIT;

df.shape[0]

PROC DATASETS LIB=WORK; MODIFY MYDATA; RENAME old_name = new_name; QUIT;

df.rename(columns={'old_name': 'new_name'}, inplace=True)

DATA WORK.MYDATA2; SET WORK.MYDATA; KEEP col1 col2 col3; RUN;

df2 = df[['col1', 'col2', 'col3']]

DATA WORK.MYDATA2; SET WORK.MYDATA; WHERE col1 > 0 AND col2 < 0; RUN;

df2 = df[(df['col1'] > 0) & (df['col2'] < 0)]

DATA WORK.MYDATA2; SET WORK.MYDATA; new_col = col1 + col2; RUN;

df['new_col'] = df['col1'] + df['col2']

PROC SORT DATA = WORK.MYDATA; BY descending col1; RUN;

df.sort_values('col1', ascending=False, inplace=True)

DATA WORK.MYDATA_TOTAL; MERGE WORK.MYDATA1 WORK.MYDATA2; BY key_col; RUN;

df_total = pd.merge(df1, df2, on='key_col')

DATA MYDATA_SQRT; SET MYDATA; SQRT_VAL = sqrt(VALUE); RUN;

df['sqrt_val'] = df['value'].apply(np.sqrt)

PROC SQL; SELECT DISTINCT COL1 FROM MYDATA; QUIT;

df['col1'].unique()

PROC SQL; SELECT COL1, COUNT(*) FROM MYDATA GROUP BY COL1; QUIT;

df['col1'].value_counts()

DATA MYDATA; SET MYDATA; IF COL1 = 'OLD' THEN COL1 = 'NEW'; RUN;

df['col1'].replace('OLD', 'NEW', inplace=True)

PROC SQL; DELETE FROM WORK.MYDATA WHERE col1 < 0; QUIT;

df = df[df['col1'] >= 0]

DATA WORK.MYDATA; SET WORK.MYDATA; DROP col1; RUN;

df.drop('col1', axis=1, inplace=True)

PROC TRANSPOSE DATA=WORK.MYDATA OUT=WORK.TRANPOSED; BY subject; VAR scores; RUN;

transposed = df.pivot(index='subject', columns='scores')

DATA MYDATA; SET MYDATA; IF COL1 > 0 THEN NEW_COL = 'POSITIVE'; ELSE IF COL1 < 0 THEN NEW_COL = 'NEGATIVE'; ELSE NEW_COL = 'NEUTRAL'; RUN;

conditions = [(df['col1'] > 0), (df['col1'] < 0)] choices = ['POSITIVE', 'NEGATIVE'] df['new_col'] = np.select(conditions, choices, default='NEUTRAL')

PROC STDIZE DATA=WORK.MYDATA OUT=WORK.NEW_MYDATA REPLEV=MISSING METHOD=MEAN; RUN;

df.fillna(df.mean(), inplace=True)

PROC GLMMOD DATA=WORK.MYDATA OUTDESIGN=WORK.DESIGN; CLASS COL1; MODEL COL2 = COL1 / SOLUTION; RUN;

design = pd.get_dummies(df, columns=['col1'])

PROC FREQ DATA=WORK.MYDATA; TABLES COL1; RUN;

pd.crosstab(index=df['col1'], columns="count")

DATA MYDATA_SUB; SET MYDATA; IF COL1 > 0 AND COL2 <= 50; RUN;

df_sub = df[(df['col1'] > 0) & (df['col2'] <= 50)]

PROC CONTENTS DATA=MYDATA OUT=COLNAMES(KEEP=NAME); RUN;

colnames = df.columns.tolist()

PROC MEANS DATA=MYDATA MEAN MIN MAX; VAR COL1; RUN;

df['col1'].describe()

PROC SQL; SELECT A, B, MEAN(C) FROM MYDATA GROUP BY A, B; QUIT;

df.groupby(['A', 'B'])['C'].mean().reset_index()

PROC SQL; SELECT A, SUM(B) FROM MYDATA GROUP BY A; QUIT;

df.groupby('A')['B'].sum().reset_index()

DATA MYDATA; SET MYDATA; COL1_NUM = INPUT(COL1, BEST.); RUN;

df['col1_num'] = df['col1'].astype(int)

DATA MYDATA; SET MYDATA; IF MISSING(COL1) THEN COL1 = 0; RUN;

df['col1'].fillna(0, inplace=True)

DATA MYDATA; SET MYDATA1 MYDATA2 MYDATA3; RUN;

df = pd.concat([df1, df2, df3])