aboutsummaryrefslogtreecommitdiff
path: root/src/describe.py
diff options
context:
space:
mode:
authorCharles <sircharlesaze@gmail.com>2020-01-25 13:06:10 +0100
committerCharles <sircharlesaze@gmail.com>2020-01-25 13:06:10 +0100
commitdea0f4cdec5bdf24962c8ab3ab2a6473e202259a (patch)
treea2a703a50b2d744e103a657d50ea793743ce1ff5 /src/describe.py
parentd5e51613d3582e18e858055cf4874507a0df452f (diff)
downloaddslr-dea0f4cdec5bdf24962c8ab3ab2a6473e202259a.tar.gz
dslr-dea0f4cdec5bdf24962c8ab3ab2a6473e202259a.tar.bz2
dslr-dea0f4cdec5bdf24962c8ab3ab2a6473e202259a.zip
Custom statistics modulde, describe program
Diffstat (limited to 'src/describe.py')
-rw-r--r--src/describe.py20
1 files changed, 20 insertions, 0 deletions
diff --git a/src/describe.py b/src/describe.py
index e69de29..7a968f1 100644
--- a/src/describe.py
+++ b/src/describe.py
@@ -0,0 +1,20 @@
+import sys
+
+import pandas as pd
+from analysis import Analysis
+
+
+if __name__ == "__main__":
+ if len(sys.argv) != 2:
+ print("Usage: {} dataset_path".format(sys.argv[0]))
+ sys.exit(1)
+ try:
+ df = pd.read_csv(sys.argv[1])
+ except FileNotFoundError:
+ print("Could not find dataset at: {}".format(sys.argv[1]))
+ sys.exit(1)
+ df = df.loc[:, 'Arithmancy':'Flying']
+ df.dropna(inplace=True)
+ a = Analysis(df)
+ a.describe()
+ print(df.describe())