From b9c75eae3b55a1cb55947e363ca488b52477503e Mon Sep 17 00:00:00 2001
From: arunabh15091989 <Arunabh.social@gmail.com>
Date: Fri, 30 Jun 2017 17:33:26 +0000
Subject: [PATCH] Done

---
 build.py                                 |  41 +++++++++++++++++++----
 build.pyc                                | Bin 0 -> 3147 bytes
 tests/__init__.pyc                       | Bin 0 -> 173 bytes
 tests/test_get_categorical_variables.pyc | Bin 0 -> 2772 bytes
 4 files changed, 34 insertions(+), 7 deletions(-)
 create mode 100644 build.pyc
 create mode 100644 tests/__init__.pyc
 create mode 100644 tests/test_get_categorical_variables.pyc

diff --git a/build.py b/build.py
index 35cdd2a..310c43b 100644
--- a/build.py
+++ b/build.py
@@ -1,26 +1,53 @@
+import pandas as pd
+import matplotlib.pyplot as plt
 def get_categorical_variables(df):
-    return []
+    result = []
+    for i in df.columns:
+        if df.loc[:,i].nunique() < 20:
+            result.append(i)
+    return result
+
 
 
 def get_numerical_variables(df):
-    return []
+    cols = df._get_numeric_data().columns
+    res=list(cols)
+    return list((x for x in res if x not in get_categorical_variables(df)))
 
 
 def get_numerical_variables_percentile(df):
-    pass
+    final = pd.DataFrame(columns = ['variable name','mean','median','25th percentile','50th percentile','75th percentile'])
+    final = df[get_numerical_variables(df)]
+    return final.describe()
 
 
 def get_categorical_variables_modes(df):
-    pass
+    x = df[get_categorical_variables(df)].mode()
+    # return x.to_dict(orient='records')
+    return x
 
 
 def get_missing_values_count(df):
-    pass
+    return pd.DataFrame(len(df.index) - df.count())
 
 
 def plot_histogram_with_numerical_values(df):
-    pass
+    df_sub = df.loc[:,cols]
+    sub_cols = list(df_sub.columns)
+    col_length = len(sub_cols)
+    for i in range(0,col_length):
+        plt.figure(figsize=(30,10))
+        plt.subplot(col_length,1,i+1)
+        plt.hist(df_sub.loc[:,sub_cols[i]], bins=10)
+        plt.title(sub_cols[i])
+        plt.xlabel("Value")
+        plt.ylabel(sub_cols[i])
+        plt.show()
 
 
 def plot_facet_box(df):
-    pass
+    df_sub = df.loc[:,get_numerical_variables(df)]
+    sub_cols = list(df_sub.columns)
+    df_sub.groupby(sub_cols).agg('sum').unstack(1).plot(kind='bar',subplots=True)
+    plt.show()
+    
diff --git a/build.pyc b/build.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..2157473b901bc8071c9085ada09e893fad3a281a
GIT binary patch
literal 3147
zcmcImU2hvj6uo1+Nt`A$MWt;(&<GV_KxtA?1Qk_<)TnRxVS`#xsaWgXu|3Xu*V&nE
zU6ee9FNqia0{@GD!4H6Q?$$|Z>7$*U$(@<qnYrhjduN(|udjB0fB1bOFMbXD{|3eX
zfGWjfBoldT=5|BIO?ljsv>`vC|AM4VHI^i8sj)2S1vOSAT~ebh>9QKDlCG$6k<%V@
z+E~K<gI_l>Fq<eXl=~?D0M$_B#K`P{9E*%a_Mdz$Ck<`X+Y$L0&4vsbvN>+ZsqXer
zq}-5W!+m1x)QCCS5Im32chI4KU3Ud_hiippR7sIlliUYxAS)8>%d6ZSRW>Z4FP=_q
zo_07K_<cIi?$Y`y3xQjA9aN%~{}$HHsF>KDxUBNHKf1lQd*}9@ukP$TD#}AYjT5_*
z6sg^@X}mos%I&1cpW4#9BHu3KV7Jpa#NLLTe&w?C*7SMU#Ee55dI?5{Md^|_>phK2
z7xy#kxhublO3aekFdvyK<_t}a>G&Ru-xS1C)P}Ia6958u04eNWMFv3C$S*?J_A!9x
z1ztKKJtt&ep!knbft4rkb1KXebbfqsxGW`rH}cRNH7S~d4gu<F9T9D%KykdatKiM*
zG0wdp1U;2@cW84vo0j)3q2cMxlGL1DhTS8;c{Q<sw?_sdbV}f+GUtQ77_p^YP)scs
zXrQK^PU`pq_`im^U)tWw?5|<SQ=?)onYEgEy0(IHAI0B81#-YSxJA$+TvJ3Z;uaCR
zEW;&)jUk7sZPeg!v*r{51hpJ=I9FHU@ar(=%G=80iB&WvHqJF7CN_0(?g{YcdttOS
zwPj-S;2=}>?d`ta_vL@ztAmmrOw&Nt_Cdvi5-UY-p%u87)cT}!eXByMIk|xDfy?8p
zCIc;501qEx+BfjfJ9qVME?!e|69IV^I}Ajn<XsqqTWi7%asYG<`aE0+frd+sF!IH@
zyAgp|sHZouz#B78Z-U+U>v+X8AjDj4sFU*?u1;=Ypw=A|4Fm`=smjJJI(r?BxvL~E
zVhl!vuFk?x4ZB^Q+L>BOQRU&4M4=EYfM>c!?<dZCmk*I^v&#COZs7xrY4VVoS?-c#
ztJ8nMps&!VJ~*C`;dnfO+&babI56k7UK~#+O_}{Ea1sG}43v<3pZp;wEkQ_)ahf7#
zR~V#*?G?U)=sngLzINIa?QZLV_tLUR;do_}OEb`cxv3{u<*FP40s1M<;$>2!jkQ-t
z*GZN++LuT0s6Mf~4qny==fDlU`_Vdr;&=u*YCsj+cQ{Z9&O?Hy+C6`*!&r_qWio+9
z7g?c5MCkRiGR{Fk6U`~kSR#zYs?zG=?5qAXD}pXGLUz_Kf(u!FUeB_)Z!`Vqd2Lds
zepEc`sMgrh(R9$mRGL)q9|R+-(+wEC%;v)|in#JZL+vIOpff!>QVfymde2-K&CeoD
z_LLu!kH3ma%%<l0t4J=_%%|qESw~6c(MJzej1w5IixE8_`~M}0QybA)Uz*0MN4z?Q
zeDoScEzS=QKz1q(gA6@hRfVWJh*eiP^VQ^B3Wtbd?d`|q9BX>)JIcGtKm~Co10{J_
z7S*)>T)&KmL+z^|syI1R;?ETDU0K;^3u{Nbm7Jf{5qXO~VIRt(?iPVtR{{sP*@a%e
znC+s+uW(Is1618M9jaUvH;wZ&_L}ViQ{E&F{5EraT=di<f>@iiPrNee*%!S-i+6$O
pJyxHw;>GUVJGB;&5b4WZWc@0$_ehuDK(z+f+U@p|{<Yd~{R`#&p?m-U

literal 0
HcmV?d00001

diff --git a/tests/__init__.pyc b/tests/__init__.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..3743001ffd7b97d4af87cf8f2f681d24e74ec8c4
GIT binary patch
literal 173
zcmZSn%**BdCNwgc0SXv_v;z<qvjB+{28Lh_kcgiKkYGR~io<|n`WgATsrrdUrFn@-
z8HT0?mWGxVmipoOMcKs#iOH$@$@wX%`l%_2x@q}Ey2<%@WvNBQnfZCTMTsS;x+#ez
siN&cU`X#BwCB^#j@tJv<CGqik1(hWnK%;GP^HWN5QtdzvC<bB%0Pcw?(*OVf

literal 0
HcmV?d00001

diff --git a/tests/test_get_categorical_variables.pyc b/tests/test_get_categorical_variables.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..970ec3230cd8f4e74316106916be3ced3e5d68bf
GIT binary patch
literal 2772
zcmc&$%Wl+26uovolI9`8V|a;&KuGi^X;>glB!mz_AcVxi0U>&0RqX1-!ELu(<&2QR
ziuv5x%?He!a~(QOl1400x|6#0EnipFIp^LgxBfkU=IuWpJ|%MSYvKAF%?)8v{0EYY
zi~|`463`DN4`tYrv?ar~*p}>e#J1%dBD(4>iS4SpEGYs483mGd<ZqF0BA<pm4zbzq
zV)*Po{Q3dz{4&gY>->gs_MN*4Z*7noZw5(O?Agj?Wf9X~OWdBneGvTG`ZzJ(ZkJV-
zm^|JyRc5ww>!xTp_6s!k1jY(M00WjX42g_4KoaUrn~>Pk;ktR#m1wiy!J7%kAv0_n
zXs(380Axei`3iRS;kU$xz!`iEWnW|$gSTZrl+VgYaM0B`-qk(cwKUU}3MeB-?`Al=
z0e^cQ1y}}FIPy!lBxPOr>Wh+G*uUc1*~(J@SJqWxeNSV^v$tu#4bKzxR-NT(#Bn0x
z_Ge&><Gb&x+V-`N!OrF*4fE`(tn1?<Y8=>+!A?2015?$7+1h#h<k9-$^*`4KAIs{q
zo0!Bx;?xdoYF0;OwR))cY6WRmDY~<M;E^0oZ9EtA+T@GBg9#ZsKZ<DGIT#66f(yZA
z{X4Gk8wC6og;hJUJ*(X{_{34vV|yfd8?s5ILBwTTed&?i6Sywk%Ut$drHC03ou^@V
zL}V&UM5&qwXW56gC_Qdr>Zu+lpuLFajyS%nF;#-8wZW%PVSY&oueFBU-9@>f44WNm
zQ!{#&{QDSi4*BuKq9O{P<yJZEvwj+$xza<PwHlyB;k+qOLoAa?s-^T2Df_B2<EBI!
zE%fC@Yw-|k{FxRr;F+I}N!!SB3ZmS_TyrZs9%GMiM-{n>R}K}4F41MZL|0&D3PIx*
zP5uL{X-fWyr&a@!1#`O0U&oAOew;bv^EQsryv9zWr^Zpr8T%`gYuutK=Xb)C@{`cr
z`;lucc0DEIxIq2Ju~vCZVd@-i#A72`q4|S`RTA+SKHOpT?$nRHh#x52MICqkZqQrq
zb$gv&r%zWA)kWsXMT8E~!+&Cm)Hp2|Y@8w^?0FhmS(C>6&SEdmU#+J`b<VSBu)4^n
UjV8~yAO*b=oC~`+&o6}k1G>tXr2qf`

literal 0
HcmV?d00001