From bc5627f9b580d7fc89e74eb7553f497ca4953e5b Mon Sep 17 00:00:00 2001
From: datanomnom <ronzzz92@gmail.com>
Date: Thu, 6 Jul 2017 06:32:23 +0000
Subject: [PATCH 1/3] Done

---
 build.py                                 |  50 +++++++++++++++++------
 build.pyc                                | Bin 0 -> 2979 bytes
 tests/__init__.pyc                       | Bin 0 -> 168 bytes
 tests/test_get_categorical_variables.pyc | Bin 0 -> 2737 bytes
 4 files changed, 38 insertions(+), 12 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..90f2e7e 100644
--- a/build.py
+++ b/build.py
@@ -1,26 +1,52 @@
-def get_categorical_variables(df):
-    return []
+import numpy as np
+import pandas as pd
+from pandas import Series, DataFrame
+import operator
+import matplotlib.pyplot as plt
 
+df = pd.read_csv('data/conversion_data.csv')
 
-def get_numerical_variables(df):
-    return []
+def get_categorical_variables(df):
+    return df[['country','source','new_user','converted']]
 
+def get_numerical_variables(df):
+    return df._get_numeric_data()
 
 def get_numerical_variables_percentile(df):
-    pass
-
+    df_temp = get_numerical_variables(df)
+    return df_temp.describe().T
 
 def get_categorical_variables_modes(df):
-    pass
-
+    dic = {'converted':0, 'country':'', 'new_user':0, 'source':''}
+    for col in df.mode().columns:
+        dic[col] = df.mode()[col][0]
+    return pd.DataFrame(dic.items(), columns=['var_name', 'mode'])
 
 def get_missing_values_count(df):
-    pass
-
+    my_ans = pd.DataFrame(df.isnull().sum(), columns=['missing_value_count'])
+    my_ans.index.name = 'var_name'
+    return my_ans
 
 def plot_histogram_with_numerical_values(df):
+    fig, axes = plt.subplots(2, 2)
+    df1 = get_numerical_variables(df)
+    list_of_cols = df1.columns
+    axes[0,0].hist(df[list_of_cols[0]])
+    axes[0,0].set_title(list_of_cols[0])
+    axes[0,1].hist(df[list_of_cols[1]])
+    axes[0,1].set_title(list_of_cols[1])
+    axes[1,0].hist(df[list_of_cols[2]])
+    axes[1,0].set_title(list_of_cols[2])
+    axes[1,1].hist(df[list_of_cols[3]])
+    axes[1,1].set_title(list_of_cols[3])
+    plt.tight_layout()
+    plt.show()
     pass
 
-
 def plot_facet_box(df):
-    pass
+    list_of_cols = df.columns
+    for col in list_of_cols:
+        plt.boxplot(df[col], 1)
+        plt.title(col)
+        plt.show()
+        pass
diff --git a/build.pyc b/build.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..0ed5dcdb82057820e649041919b3ca6858c7d9dd
GIT binary patch
literal 2979
zcmcIm+iDy~5bd75NV}3O*|8NbI1V_5ECgwB2!enK0gI5l?87J^ve*ox>E6{Sb6HRK
z>H^Y}^B4KDd`F%`a!!r*;t=P_lGLT@?wanZ)2FJt@#j|e^&fA3N!0C6fdAj4x!)lS
z{<O-J8tJ(ns8K`oKxKi-n`+clc}tC2O5a+y)u=tUUQ=i5YSdA-q0YL>G}Jq-MjOgB
z)o)mIQ<;`1TgtRWxu?vUDBH@ci?X9kN0j@@bVYeUpN{$)@Gt!vKf^wK_HBqGTcy_d
z4%EZg$3IkYZrx+_40?m4EY5A^(z1xy-%s3mqA|(VLVJbg9ztA8*e0OErl}@P)jR6b
zV(8Z(lX6!0>PmKW<*Z7qU&nA^FQS>Vm26+{<*n&sM)kRcr<-P64{+4|2r)P<b333v
zMVaGoa9mcux@nwP_-pLIns{$qR(nh5_o~?2J$~V=ADqn6%<NCEd=HCFtdA1RPRc4x
z;w(Ckt292ztRo_h`zyT&nLbI#T__0oOsRK(P<$QM9^c1I#6^o)4rmj>O{PM^eOIA;
zinUjET@7RtQ>%e6CN>F#5O|3NG`rF2TB|e7pw5CtTn*TSghS5ESeI1kiS-2IXCabA
zCg*I%k+=EuuA+Q~Ek6)tG__z=;nU1+!^$-wl-{h*ju9;A>d;A>%m~o+pyulVd#$?H
zRtEs$a<DvLAaeNTsk#Z4L)U@o;hP@tg8K69*XkH#J!=SnO#+ClL_Q)DCjnAeAN9$Z
zVG5x_`U2FU@gPF_8q&wxT*+p6;RI~3F)EOeo+dfSEjbmEWFf<p1euypl7*|LfZz4y
z#7H+!&7_I^Ps;4B4(4-g^TAAt=(zg?CT<-<>1TRZKdDcK@iVqpXzm*bzyi3LTYwD`
z^>1EUCLpHPI2)W9dD>J_i^@@-D-3HuxJ{lqmlhM`ayGM()Shq$<9$LBKA;f5*iBtA
z%QC^o&2ldo7RFu*;)H&M;k3`MqPTE(g(C|uaeOZVSBJn<b#y`L=W}Ay-35Gqp;N<0
zHUq3cy$+bvATEdkOqt`r4mV3f-89KukiG{XF?WrW0I7Sh(VS>w0`S(O^TD-M>&A-p
z`obEcbL-BkHIij9(Op<$bZ)(|YK{A2F|oO@#we{BWShJYZA~=8I=lr0hABBEd6x2&
z(*}ebVm*f5|8)7qit(~5OD(_sukOE$rAxvKxu%(yjCQk=X;ykC1+}Iu5AZZ~UT#5L
zP?1l4W<Be+Pba57%HpeX=7o{&w7jTm0GS{wkJE`v#h19p8LSz<kdHI08I@zOC%ZGk
zKF5BHCt?8Q1h0qaBK4>9+fVKxB%yl>JNE=a>Bk^%NB8uj_gYtXm$iC9>7VccYk?w-
zFTZ8BAHI1Zghl}nPJ^qg#2~xImBb3ndRup-5bL))>R9MOVhn{8HH567Aw842ds1F9
z5-9<40^t{!TGrC=OPW987+Nhp?tA1OmK89^c+2<*4((&akq8QqZRv+-UEODm6F2e3
zz%zzn6;pXzP2<AEt}b!>s;+Y+VTt=O_GRUHcjU3>K3RH#XOIL<Hq~kk2tTQ8Y$Ck)
z<a!VJjH^k@EDLujKBFKS;a3n#SVa4{*wCSuxJYK1{f?O>Z`4gtX%}rf*bX`k{5Cu7
IPOG!?596M12LJ#7

literal 0
HcmV?d00001

diff --git a/tests/__init__.pyc b/tests/__init__.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..de27e4620161c7bd87decfb51f74c962cdb5b3c7
GIT binary patch
literal 168
zcmZ9FK?(vf3`Hxt5W#!Qrse`7o<JBx23MuD5sS7NnmFKn-MaDurr^r({|kJ`&-}Aq
zz22vDFu$4U-m7_`h{hbdmK-{T`2+Ww!Id&X^x_L;6%15wJo<nTC?a6=_CnPLDic;t
jRBZKvwp}m^2(mCZmr4@nHse#4+PcjYA9vif->f!Y@W(0(

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..47c1104e65660824314a960b5f76c68e1d53fb83
GIT binary patch
literal 2737
zcmc&$O>fgc5FI-oZTJv{@-03DLL!$oCvFu&6oI&Ka6yP%EMsqqTgQ&P>!3>Qsr*!~
zocICo-ndOd+aM08apLjLdc5BE-psBWKQCYS{^P@^SdNYct}k%fK1_nYCz;65lfEY&
zeP1$P`VC1M(r=1s$WBX4Qw|Z)R(DBETis<z5a7wclcXizL=Hth^*bD5tJ}u#=`S2z
z4=uM0^WIpu8Cmnr-i5c`OClHb;vyfL(xyco(qE75*xf%0eq&r1N6u^)Wg16WIF8CR
z`kWa%LBp|M;<V3Tj1UAcU@Co|$k+f9UmHzAVo!_nW}_{^R=0&GV~#^+cy8jf1q=ot
z8_LcB*x85Q5a$D@ci_vu$PNZ?%DykV%7}N=)jHkPKHar6)0GG)BS-IMIJ*OXdmaUt
z24*;LOSr^EmAi6JNzTpJu(GCf6u=f$85`Hp81n2)(rv<XME$u+vn1d+0dczvFp=f6
z@5{<`wU5EpWCIOz>?)1z;v#D7Hd?(e#n7mlc`?Ml_pvB<?I?;3)J;s!B+=TSDA%S^
zua!`Bje1+-dJcJE)rNE7u8;QI1H6l|^P>pVUV;&C#k=CI>hH9;ZxQfc#8ufy^sH>R
z;1fq)4b6$Xt$8Mk5h7M`b%jG>kKo#HoZ9qQpn%B`T&7`oKxC>&K$Yqn7ukn%0aDz<
zTPISS0rV>FTS7P%Vj=<)YLj<2VE#!2rxk{b-9eF|`090POY?Y^_s1A;4)5W}pbm1E
zW=84kvTACIInyb{S_6<Fw=T+2IV_G*q6PE{+4{PShIL`oGU&q#l;J6+_cs}4a5F0{
zQ?*v)9Q1gIcg?NmaEO(`o)qII9+`>}T%*e>39iG;q`$^3koPB;&V=_fcc>a4%jA4X
zzl}GN^kHg|#oO32vkFU#?hq&0X6(PPt#J!vo1X&{w$B3g=ueupx^*)Q!yE-0hFaDk
zHK{VV0k?=?h2{ngYa`%pJ6%n+b)LIhz)ujan3fBC+v_ZM+MQOX)upRosyub1AwY-d
z_CAX8B(ho|SR_RTSl=|XG*J@rONzbRX0?E7t@)k>;?((V)Ji<(Tomt$cgb&K$G+nK
E20c%fkpKVy

literal 0
HcmV?d00001


From af355ff19cfa039ff73f88067526815e4499171a Mon Sep 17 00:00:00 2001
From: datanomnom <ronzzz92@gmail.com>
Date: Sun, 9 Jul 2017 10:14:16 +0000
Subject: [PATCH 2/3] Done

---
 build.py  |  68 +++++++++++++++++++++++++++++-------------------------
 build.pyc | Bin 2979 -> 3192 bytes
 2 files changed, 36 insertions(+), 32 deletions(-)

diff --git a/build.py b/build.py
index 90f2e7e..9433509 100644
--- a/build.py
+++ b/build.py
@@ -1,52 +1,56 @@
 import numpy as np
 import pandas as pd
-from pandas import Series, DataFrame
-import operator
+from scipy.stats import norm
+import seaborn as sns
 import matplotlib.pyplot as plt
 
-df = pd.read_csv('data/conversion_data.csv')
+df = pd.read_csv("data/conversion_data.csv")
 
 def get_categorical_variables(df):
-    return df[['country','source','new_user','converted']]
+    return df[['country','new_user','source','converted']]
 
 def get_numerical_variables(df):
-    return df._get_numeric_data()
+    numerical = df._get_numeric_data().columns
+    return list(numerical)
 
 def get_numerical_variables_percentile(df):
-    df_temp = get_numerical_variables(df)
-    return df_temp.describe().T
+    final = df[get_numerical_variables(df)]
+    return final.describe()
 
 def get_categorical_variables_modes(df):
-    dic = {'converted':0, 'country':'', 'new_user':0, 'source':''}
-    for col in df.mode().columns:
-        dic[col] = df.mode()[col][0]
-    return pd.DataFrame(dic.items(), columns=['var_name', 'mode'])
+    modes = get_categorical_variables(df).mode()
+    return modes
 
 def get_missing_values_count(df):
-    my_ans = pd.DataFrame(df.isnull().sum(), columns=['missing_value_count'])
-    my_ans.index.name = 'var_name'
-    return my_ans
+    df1 = df.apply(lambda x: sum(x.isnull()), axis=0)
+    return pd.DataFrame(df1)
+
 
 def plot_histogram_with_numerical_values(df):
-    fig, axes = plt.subplots(2, 2)
-    df1 = get_numerical_variables(df)
-    list_of_cols = df1.columns
-    axes[0,0].hist(df[list_of_cols[0]])
-    axes[0,0].set_title(list_of_cols[0])
-    axes[0,1].hist(df[list_of_cols[1]])
-    axes[0,1].set_title(list_of_cols[1])
-    axes[1,0].hist(df[list_of_cols[2]])
-    axes[1,0].set_title(list_of_cols[2])
-    axes[1,1].hist(df[list_of_cols[3]])
-    axes[1,1].set_title(list_of_cols[3])
+    num_cols = get_numerical_variables(df)
+    plt.figure(figsize=(15,6))
+    plt.subplot(121)
+    plt.title(num_cols[0])
+    sns.distplot(df[num_cols[0]], color='yellow', fit=norm, kde=False)
+    plt.subplot(122)
+    plt.title(num_cols[1])
+    sns.distplot(df[num_cols[1]], color='yellow', fit=norm, kde=False)
+    plt.subplot(211)
+    plt.title(num_cols[2])
+    sns.distplot(df[num_cols[2]], color='yellow', fit=norm, kde=False)
+    plt.subplot(212)
+    plt.title(num_cols[3])
+    sns.distplot(df[num_cols[3]], color='yellow', fit=norm, kde=False)
     plt.tight_layout()
     plt.show()
-    pass
 
 def plot_facet_box(df):
-    list_of_cols = df.columns
-    for col in list_of_cols:
-        plt.boxplot(df[col], 1)
-        plt.title(col)
-        plt.show()
-        pass
+    plt.figure(figsize=(10,10))
+    plt.subplot(221)
+    plt.title('Age')
+    sns.boxplot('converted','age',data=df)
+    plt.subplot(222)
+    plt.title('Total Pages Visited')
+    sns.boxplot('converted','total_pages_visited',data=df)
+    plt.tight_layout()
+    plt.show()
diff --git a/build.pyc b/build.pyc
index 0ed5dcdb82057820e649041919b3ca6858c7d9dd..aa64a6a9c0620345473ea039e3e9f51619cad6e5 100644
GIT binary patch
literal 3192
zcmcIm-EJF26rS}@5<5Rm(yBE4h)_Tnq15VqRe?ZNuOj3i5V={cJ!5Cm@vgNqo77dg
z3KzTpkHizeExZH|0N;1UiQOvn3Mbi<nX|KJ&Uemt&P@I9?e=fU;K!kw|J3mROFaHp
z2#vo)Ii*gM>v2t;)@58%uAwGPb=p+PeBGKlUB6zpq0U?Cw5?2Coo^~#S67KT-BP-t
zeulYirJFL?QF=`V9i`W0u&eZj4EB_6$zWgUwhZpjo0IM)JhT7c=eV0-7#a|HQBM4Q
zx}1fqH!SiCQ+iwEDd!J|{$j|be42Q^#^dioyb^%AMhA66jT`FVq{|h@VI6W<RCy?8
zVFTmbT&9&bWmv<+7gaelp(Sg$S1`JZiHaUEh<X$i)cR)-y|ZFsdh{nRCiv_9P?XR8
zG#eWDt4+`7>}XV!M+@hV$}E^8e&J2%4JzyO!|5y>z}VP?bcna(qO`-zr59Ogvw<_7
zE_v>+5`oMiAA!s-0r_01E67Csl&DZsuM#HbL|i}a*0Cn+L8ROxuO<dNNEqiKan&UE
z5_D&M=+bzfA(Lp#*Jo~3;NQVcOO7o=-@+?TGN`1p08cyu{#^_JK8OV11E2?VgaN_0
zMlWbcFv+<#epuRpSprF<Bb%=d-~(*%9{|&-0qgQ$o!P~zIo90;)|*7%BZ6z75n)98
z8cs4<5;+0?-oQweUsV!6oMXI|!KrM`>YM}IUjSrwC7GXM5SfgYfQ}MN_$o<AX(V;Q
zYJPHjQ~iw42_DY?#t6mmEL8y41k%P=6RCFV^UAqc+H4^&Ruz{G7=P@t$v|gMSgRgS
z%ux2iddpd?PqHwZPTdUUtJG9W?R|qr@wCh)%c&#`J$ke%huLvS)`|7r=3}%WR~es5
z-{iY_2JbI2NQjcbXNcqtMQ^f(Z0cPUWRft8(v*=InKW3fI5rMM(NyPaWWZc^fMWy)
z2#yYPDV!T9->am4ZKTZxw`64V79-HT-N@D(Mz(J;0^Qq<?7U&5bBht^-eRQ1b|qMV
z3M0+t#eN-+tdSji`_kABG?oQ!XB2o;AZvz&E8-b8GtL#4LiCXh^8dLu_9d5-<@R?9
z^2`1~K`Kd;SlE`MxG(LJZ>KIuy}*a6G;&<{YA|(0kgbCa&cwpP0YV~Z93Xt%$2hw8
zG;=uX*vW$(pM}(Av!V*pr2JWNxzwciJx&{N&8@6YeS$scUwQ^di!o{|y|m%%`c#y|
zhlKwC>--^vO16^SWINeQKD$M>V;Q`53;6{fp$yQ3Nd3)7eaz)bBCN$^Q3nlm-eeI(
zx`Y2$HIzzRPs)QCaXo0l$<6vU%VTN%TBH#cuP<oCxF7f5hqP^qKj|DZ`_<U=*+^xQ
zq2f8=#l-%FfxiRsT@f;O_$^-g!|&0oa3c!T2kO%)_31^_M;6Hr6kVpW{}IIlijOH6
z`aZk7&}dM+5T10Y{TxD?v0OW-QX>A}Th<QDO72Ivp@kG4^q}$k+)i@5o!mo?Dcec+
zF0?{&P!d-(R!u&YVwh&R&U~NUFtAem&`xI$eaM23wC6?Bn`}^&c?8U2o@7!O&JJ+n
zqZ`qd3?)=jKc&fZip#A8qR(oVV(ioRKF##sqxgd25yT=^D+t&IHh7G)yK?3UDK6LX
VEh&htq*LqETJ=^#ew(fRe*tuDu0H?(

literal 2979
zcmcIm+iDy~5bd75NV}3O*|8NbI1V_5ECgwB2!enK0gI5l?87J^ve*ox>E6{Sb6HRK
z>H^Y}^B4KDd`F%`a!!r*;t=P_lGLT@?wanZ)2FJt@#j|e^&fA3N!0C6fdAj4x!)lS
z{<O-J8tJ(ns8K`oKxKi-n`+clc}tC2O5a+y)u=tUUQ=i5YSdA-q0YL>G}Jq-MjOgB
z)o)mIQ<;`1TgtRWxu?vUDBH@ci?X9kN0j@@bVYeUpN{$)@Gt!vKf^wK_HBqGTcy_d
z4%EZg$3IkYZrx+_40?m4EY5A^(z1xy-%s3mqA|(VLVJbg9ztA8*e0OErl}@P)jR6b
zV(8Z(lX6!0>PmKW<*Z7qU&nA^FQS>Vm26+{<*n&sM)kRcr<-P64{+4|2r)P<b333v
zMVaGoa9mcux@nwP_-pLIns{$qR(nh5_o~?2J$~V=ADqn6%<NCEd=HCFtdA1RPRc4x
z;w(Ckt292ztRo_h`zyT&nLbI#T__0oOsRK(P<$QM9^c1I#6^o)4rmj>O{PM^eOIA;
zinUjET@7RtQ>%e6CN>F#5O|3NG`rF2TB|e7pw5CtTn*TSghS5ESeI1kiS-2IXCabA
zCg*I%k+=EuuA+Q~Ek6)tG__z=;nU1+!^$-wl-{h*ju9;A>d;A>%m~o+pyulVd#$?H
zRtEs$a<DvLAaeNTsk#Z4L)U@o;hP@tg8K69*XkH#J!=SnO#+ClL_Q)DCjnAeAN9$Z
zVG5x_`U2FU@gPF_8q&wxT*+p6;RI~3F)EOeo+dfSEjbmEWFf<p1euypl7*|LfZz4y
z#7H+!&7_I^Ps;4B4(4-g^TAAt=(zg?CT<-<>1TRZKdDcK@iVqpXzm*bzyi3LTYwD`
z^>1EUCLpHPI2)W9dD>J_i^@@-D-3HuxJ{lqmlhM`ayGM()Shq$<9$LBKA;f5*iBtA
z%QC^o&2ldo7RFu*;)H&M;k3`MqPTE(g(C|uaeOZVSBJn<b#y`L=W}Ay-35Gqp;N<0
zHUq3cy$+bvATEdkOqt`r4mV3f-89KukiG{XF?WrW0I7Sh(VS>w0`S(O^TD-M>&A-p
z`obEcbL-BkHIij9(Op<$bZ)(|YK{A2F|oO@#we{BWShJYZA~=8I=lr0hABBEd6x2&
z(*}ebVm*f5|8)7qit(~5OD(_sukOE$rAxvKxu%(yjCQk=X;ykC1+}Iu5AZZ~UT#5L
zP?1l4W<Be+Pba57%HpeX=7o{&w7jTm0GS{wkJE`v#h19p8LSz<kdHI08I@zOC%ZGk
zKF5BHCt?8Q1h0qaBK4>9+fVKxB%yl>JNE=a>Bk^%NB8uj_gYtXm$iC9>7VccYk?w-
zFTZ8BAHI1Zghl}nPJ^qg#2~xImBb3ndRup-5bL))>R9MOVhn{8HH567Aw842ds1F9
z5-9<40^t{!TGrC=OPW987+Nhp?tA1OmK89^c+2<*4((&akq8QqZRv+-UEODm6F2e3
zz%zzn6;pXzP2<AEt}b!>s;+Y+VTt=O_GRUHcjU3>K3RH#XOIL<Hq~kk2tTQ8Y$Ck)
z<a!VJjH^k@EDLujKBFKS;a3n#SVa4{*wCSuxJYK1{f?O>Z`4gtX%}rf*bX`k{5Cu7
IPOG!?596M12LJ#7


From 298ca5daa4d9dfec07e6131f559cbd120aaefe53 Mon Sep 17 00:00:00 2001
From: datanomnom <ronzzz92@gmail.com>
Date: Sun, 9 Jul 2017 13:24:24 +0000
Subject: [PATCH 3/3] Done

---
 build.py  |   8 ++++----
 build.pyc | Bin 3192 -> 3192 bytes
 2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/build.py b/build.py
index 9433509..63a0c8a 100644
--- a/build.py
+++ b/build.py
@@ -29,16 +29,16 @@ def get_missing_values_count(df):
 def plot_histogram_with_numerical_values(df):
     num_cols = get_numerical_variables(df)
     plt.figure(figsize=(15,6))
-    plt.subplot(121)
+    plt.subplot(221)
     plt.title(num_cols[0])
     sns.distplot(df[num_cols[0]], color='yellow', fit=norm, kde=False)
-    plt.subplot(122)
+    plt.subplot(222)
     plt.title(num_cols[1])
     sns.distplot(df[num_cols[1]], color='yellow', fit=norm, kde=False)
-    plt.subplot(211)
+    plt.subplot(223)
     plt.title(num_cols[2])
     sns.distplot(df[num_cols[2]], color='yellow', fit=norm, kde=False)
-    plt.subplot(212)
+    plt.subplot(224)
     plt.title(num_cols[3])
     sns.distplot(df[num_cols[3]], color='yellow', fit=norm, kde=False)
     plt.tight_layout()
diff --git a/build.pyc b/build.pyc
index aa64a6a9c0620345473ea039e3e9f51619cad6e5..4834f7f9d0fcfac2f394dc78b87f185a50870b42 100644
GIT binary patch
delta 48
xcmew%@k4@*`7<w9rQSxiyX=g2CqH4=6TQd4z>vuZBJP6-CJ^ypvpPp03jlhm4Uhl;

delta 48
xcmew%@k4@*`7<xqQO1pIci9;$CqH4=6Rl!kV8~<y5tl&(6NtF7S)C)01psH@49@@n