From 6155cce4dab038e4ec77d83eae098932bc7bf64c Mon Sep 17 00:00:00 2001
From: rianboy18 <pranavr95.pn@gmail.com>
Date: Sat, 29 Sep 2018 06:21:54 +0000
Subject: [PATCH 1/6] Done

---
 __pycache__/__init__.cpython-36.pyc           | Bin 0 -> 153 bytes
 .../__pycache__/__init__.cpython-36.pyc       | Bin 0 -> 171 bytes
 .../__pycache__/build.cpython-36.pyc          | Bin 0 -> 755 bytes
 q01_missing_value/build.py                    |  25 ++++++++++++-
 .../tests/__pycache__/__init__.cpython-36.pyc | Bin 0 -> 177 bytes
 .../test_q01_imputation.cpython-36.pyc        | Bin 0 -> 2326 bytes
 .../__pycache__/__init__.cpython-36.pyc       | Bin 0 -> 173 bytes
 .../__pycache__/build.cpython-36.pyc          | Bin 0 -> 805 bytes
 q02_outlier_removal/build.py                  |  33 +++++++++++++++---
 .../tests/__pycache__/__init__.cpython-36.pyc | Bin 0 -> 179 bytes
 .../test_q02_outlier_removal.cpython-36.pyc   | Bin 0 -> 1789 bytes
 11 files changed, 53 insertions(+), 5 deletions(-)
 create mode 100644 __pycache__/__init__.cpython-36.pyc
 create mode 100644 q01_missing_value/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q01_missing_value/__pycache__/build.cpython-36.pyc
 create mode 100644 q01_missing_value/tests/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q01_missing_value/tests/__pycache__/test_q01_imputation.cpython-36.pyc
 create mode 100644 q02_outlier_removal/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q02_outlier_removal/__pycache__/build.cpython-36.pyc
 create mode 100644 q02_outlier_removal/tests/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q02_outlier_removal/tests/__pycache__/test_q02_outlier_removal.cpython-36.pyc

diff --git a/__pycache__/__init__.cpython-36.pyc b/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..af63a6f3bbfe476dc79ebba264131d4e65b38c11
GIT binary patch
literal 153
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnFE{;+{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{rLFI
cyv&mLc)fzkTO2mI`6;D2sdgaqi-DK{0CS)w>Hq)$

literal 0
HcmV?d00001

diff --git a/q01_missing_value/__pycache__/__init__.cpython-36.pyc b/q01_missing_value/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..354530f75a902270f9869d1dbb6279634ba16b24
GIT binary patch
literal 171
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuL%8&{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzr7
s_}t9mVvtx_VoqtQetdjpUS>&ryk0@&Ee@O9{FKt1R6CH3#X!se0I9$&(f|Me

literal 0
HcmV?d00001

diff --git a/q01_missing_value/__pycache__/build.cpython-36.pyc b/q01_missing_value/__pycache__/build.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..eff507292b32c351396b0f155848f86947f831fa
GIT binary patch
literal 755
zcmY*XJCD;q5Z?9Mz9c#w1St)DnwS#}#pwhBaY8zxlfpwXMYQLgB+l7uGiyf*f-1E9
z5Gwx9wp0}S0V-yblQ>qI*X}ppJhYGd{p9!P$A_WM*iUxn1mr&@(O0A-le}aZ-&-eh
z$vUN1`I*ldxa&Z=JD!CQN>BPSkfDs0{)lCfjAinbWiiAM5+Z?w6Vr^1kA4v=n~u2&
zUsX-l0)84u(TcQ1QK(6-Hzs*4^gD&m5QK>@@UqzK-RQDzXDiWwNiGD6CA`^^AJ0Sy
zS6JkbbIVM}A@N9_ljxFEOQoeF-EYnp{+X|N>qu|OM{LJAquBiyAG|Y0md03nx7Ob=
z8CblH=`~n~7P-GU_=Mi^g+l{|Q{zs>)OZyLHFnIPDzq-ta=sB|2Re&1P1?e8i!QZ>
zcH3w>A4g~#H}0ZnO>b3q`y{&>6H|uLiyA8vEQ+#Jg4k;b#%ZL9Ul8OaiVCP$qPFwf
z)GR8hW^SAK&hQ>F|9P9P>IzaUgu1S`r%%%NHQwk(<dEjXmM%cF9l{*cvQPkMn{KdP
zL*AwzADzzs8zjB%ic(IRtx2d(*NV2NRiDb}gcJ*V{Fp!B4>=Ag8Xp;7brpdf)tI0W
zN(zm~^pXwrO$h?kq(Pu!Iqa8pSVcrXK+fq);0ZyoXO*QhJl{`rN8PvA(HU)}E8*O3
SRu4!;JmNhkaZbFb7xBMQ(8!el

literal 0
HcmV?d00001

diff --git a/q01_missing_value/build.py b/q01_missing_value/build.py
index 7dc4d18..1edbf02 100644
--- a/q01_missing_value/build.py
+++ b/q01_missing_value/build.py
@@ -1,6 +1,8 @@
+# %load q01_missing_value/build.py
 # Default imports
+import numpy as np
 import pandas as pd
-
+from sklearn.preprocessing import Imputer
 # Data loading
 ny_housing = pd.read_csv('data/train.csv')
 # Selecting 4 most relevant variables along with target variable from the dataset fot the Cleaning and Preprocessing.
@@ -8,3 +10,24 @@
 
 
 # Write your code here:
+def imputation(ny_housing):
+#     mean=housing_data['MasVnrArea'].loc[housing_data['MasVnrArea'].notnull()].mean()
+#     housing_data['MasVnrArea']=housing_data['MasVnrArea'].fillna(mean)
+#     highly_occured=housing_data['GarageType'].loc[housing_data['GarageType'].notnull()].value_counts().index[0]
+#     housing_data['GarageType']=housing_data['GarageType'].fillna(highly_occured)
+#     return housing_data[['MasVnrArea','GarageType']]
+
+    imp_mean = Imputer(missing_values = 'NaN', strategy='mean')
+    imp_mean.fit(housing_data[['MasVnrArea']])
+    housing_data[['MasVnrArea']] = imp_mean.transform(housing_data[['MasVnrArea']])
+    housing_data['GarageType'] =  housing_data['GarageType'] .fillna( housing_data['GarageType'].mode()[0])
+    return pd.DataFrame(housing_data['MasVnrArea']),pd.DataFrame(housing_data['GarageType'])
+c=imputation(ny_housing)
+c
+
+
+
+
+
+
+
diff --git a/q01_missing_value/tests/__pycache__/__init__.cpython-36.pyc b/q01_missing_value/tests/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..cca5f8f024461d660d7c66458ccc75b67888687e
GIT binary patch
literal 177
zcmXr!<>lh_S{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuQ>gT{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzr7
y_}t9mVvtx_VoqtQeo1O^NwI!>d}dx|NqoFsLFFwDo80`A(wtN~kiEq~%m4u3F)%;?

literal 0
HcmV?d00001

diff --git a/q01_missing_value/tests/__pycache__/test_q01_imputation.cpython-36.pyc b/q01_missing_value/tests/__pycache__/test_q01_imputation.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..689bc9093d3b209f775aeb203eab2bdcf3c94b9a
GIT binary patch
literal 2326
zcmcIlTaOzx6t?G*Om0hAO6hVbgJ4BRsy6KM0#b#FO1oNAQp8pu4G&Xy>|`fQJgMzj
zk*GYS{{`_=_zQmJY5xLGoHLUq+0{xc0uuSyV;^6>&)+%zwAX9*zW(*c-y4MdK{kHy
zP3}R_91JI%=43)+O0n(aPV91*d$~99W54S2^TwnZHz^VBIN*&1ja#C{n>^qx-sYW=
zzeD1-@I@O&I){XJdG96RJwX?4+~v+OiMGyQClXN`JQ3R5XIc!-oVur-O{S$`CMzV2
zZ;gZ*mU+(9NKZvN-{Q=$zELcbd#OH++utz#y;NT+!Q$4TdX$}3&EQdCj!)QB#O*_-
z*hoB?LA!O#a`COoQjy~JuYS1kJt+Dv7(rsnN$haSotLDB4z<U9-hkRim{2zmD%4Hh
z;a#W$-a}Y+3-KJS5Tq2A#2@YvK)DY^AAm8$&;|X4z93`AxGyNDJ7fWS0@WMeL+1ur
zxJWvz?hO<!cZO#E{)3-^4<mR8)RYq;jShGDo$xR_6%tyJr+n~J9vItUS_@?!JTF;p
z-CW4Xv7Q$B(0WMhZ$x#$?a!=#QcOf&WlTORX8WJ@e<;*XddgDKPYW*kL%~d`L?YxU
zlR~IWj*_V=#&C}Q^H27ZNv1XSowB?XeFIGB$|g9tzSh0z%wEUQjlKlW(St*1JTP$r
zI{)xIm3x=XIL9x}toXeJjZJ=Qc2UeLF5SVI3tscjS2S`lWMyu`>hL<`MJNjsPMAqg
z!cj%)d5_3l6RhxRBS*ENQXMqNXKfIay?OEQ%eC}2jMk7TK>qk|AitWmf{jvWLaAWi
zh6cz9goR=<&e$D$V;|(ysCyQ4LriWhrupMF0T5ZCfj9|=g$k8`tClqzVWz{UQzkjn
z;aA}GP_c=4dMBI}C8$p<ec|TAT^%*;wOVaQ`bcWSWGbu+mv5aZw_aK(Vf`U4U|XxT
ztcUwXv?lNZcdu34LEKlflKr=kGE7;7Zc~TOZ(T;(+Wo)b3Q7JSxKgjfx(Y?0-avC5
z&6{X$ph2itW98r-aJ@P!2P;-EF|7jUBls|o|98lkE+8?D-36&qlm{UMsc98H%T)Am
z9)7o}leCZqGP4c`Y_JTZ$gkkpMy4g?aoecl%eqX?Hp3=v*DGQ2xEMV<l>#KI6N^sP
zm-_*tt}+H_VSf8EC0sV`-=jx8Tphi78{8FMk$M-+M!JYQYux)+27a&$%lvcT2ZN|(
zgCvm<eo126Nir#TnPa<~B=|biBWepzc@51?G&^YCLxZVx<CImXVun)~kocw11T^sB
z?*)y(t45-p4N93Ayhw$TiyL)5sbp{cXJwZ2`mJh$e(+_hvTGc`!_JfHrI)khq$u%$
miMz`NDuLDEgZp20FdN*hF@AyDY6RAWRNJLpr|oRh!1)UXv{S19

literal 0
HcmV?d00001

diff --git a/q02_outlier_removal/__pycache__/__init__.cpython-36.pyc b/q02_outlier_removal/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..7d88578d48c262fea61a67416e73be1db0a8d662
GIT binary patch
literal 173
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuPFVD{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzqy
w`25n6oXpgs_@dO@{IbLx{rLFIyv&mLc)fzkTO2mI`6;D2sdgYci-DK{01`GXxBvhE

literal 0
HcmV?d00001

diff --git a/q02_outlier_removal/__pycache__/build.cpython-36.pyc b/q02_outlier_removal/__pycache__/build.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..541cd33bdbd43e48aa4b85f0f4b6dd760e6c631e
GIT binary patch
literal 805
zcmZ8fyN(kv6t(A_Nj72Apioh4lV*oUv&9O5D7IN5q##xrijgzENhWdZ!7~#PlvLqo
zKzsn-;ucZx3shWZgIKL4-+O$0eC|VjesU7;etrKi@(KA#?yLauOWf564JVumlG1~B
zQupA!g`WmM7a@1{G>stQF88?410K%&0ZCgt;;k<vjUk2>GTP9ls52(R_%~{j(dRB_
zHA`yE3VEJaYZJd=)qAO5X<(*xrLT+i0Y=xV9?w|`Cca{t&EVYzacj&3ywybxIkj@Q
zT-?|Pd5ydJhGvWT24wNi(>=kHvnM-xym7a5f54ZP;~crx-*UY@S@_%?keze9cemKB
z#rDM_*xGmAfN+259Z|5v!J&e`r@Q|gtiy8Z&{o9nad0N#NIO`uw)Jb<kha|vKiX)@
zhlC6x;|i6VP^wxsLKtraOqyU?2qBs2teI#a%T*)m%J_xk@X2^wtI{}pI&@8Eu9~Wl
zv&_z1w@^Fn&WqVxY%-PSjRyXe=m!}7bCb-~3KCr~IZ>OZ7s-v%%c^8KB)Q^{Oo7$N
z&p^%!381ZLsnr7VI{EnIS*Ds=6reK=D}_}hlco^-yxf@Hy_*hZtGZ|ieMI~8G3|$E
z^vuzHj0`&_D4FD}GET|UumUzwTBdF#D|5PHbtx34nV_ykNWS%MDQXkp2jUt2C~ehs
m&l)RgXntt%jybjA=yIiaBjAM{`JtkSMl^O}r|(9+i2enN4A8Iu

literal 0
HcmV?d00001

diff --git a/q02_outlier_removal/build.py b/q02_outlier_removal/build.py
index 74df5f2..0f266b4 100644
--- a/q02_outlier_removal/build.py
+++ b/q02_outlier_removal/build.py
@@ -1,10 +1,35 @@
+# %load q02_outlier_removal/build.py
 # Default imports
 import pandas as pd
-
+import seaborn as sns
+import matplotlib.pyplot as plt
 # Data
-ny_housing = pd.read_csv('data/train.csv')
+df = pd.read_csv('data/train.csv')
 # Selecting 4 most relevant variables from the dataset fot the Cleaning and Preprocessing.
-housing_data = ny_housing[['MasVnrArea', 'GrLivArea', 'LotShape', 'GarageType', 'SalePrice']]
-
+housing_data = df[['MasVnrArea', 'GrLivArea', 'LotShape', 'GarageType', 'SalePrice']]
 
 # Write your code here:
+def outlier_removal(housing_data):
+    mean=housing_data['MasVnrArea'].loc[housing_data['MasVnrArea'].notnull()].mean()
+    housing_data['MasVnrArea']=housing_data['MasVnrArea'].fillna(mean)
+    highly_occured=housing_data['GarageType'].loc[housing_data['GarageType'].notnull()].value_counts().index[0]
+    housing_data['GarageType']=housing_data['GarageType'].fillna(highly_occured)
+
+    
+#     housing_data['MasVnrArea']=housing_data.drop(housing_data['MasVnrArea']>456.0)
+#     housing_data['GrLivArea']=housing_data.drop(housing_data['GrLivArea']>2466.1)
+#     housing_data['SalePrice']=housing_data.drop(housing_data['SalePrice']>326100.0)
+#     return housing_data
+    housing_data = housing_data.drop(housing_data[(housing_data['GrLivArea']>3000) & (housing_data['GrLivArea']<6000)].index)
+    housing_data = housing_data.drop(housing_data[(housing_data['MasVnrArea']>3000) & (df['MasVnrArea']<6000)].index)
+    
+    return housing_data
+c=outlier_removal(housing_data)
+c
+
+
+
+
+
+
+
diff --git a/q02_outlier_removal/tests/__pycache__/__init__.cpython-36.pyc b/q02_outlier_removal/tests/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..f6d2f0d0cb4a4d42f93f504f05a23b1f1590ae13
GIT binary patch
literal 179
zcmXr!<>iu`ygZr#2p)q77+?f49Dul(1xTbY1T$zd`mJOr0tq9CUkUmd`MIh3MVX0t
zN%@tA7W(1&McKs#iOH$@$@wX%`e~_&C8b5F@u_*~nR%(HMVWc&@dZWsS*gh-`h^BY
z@%g1CIhm<N@kOb*`DKYY`X#BwCB^#j@tJv<CGqik1(mlrY;yBcN^?@}KsFZxF#`aH
C^)Y(@

literal 0
HcmV?d00001

diff --git a/q02_outlier_removal/tests/__pycache__/test_q02_outlier_removal.cpython-36.pyc b/q02_outlier_removal/tests/__pycache__/test_q02_outlier_removal.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..aeebef96e456bfe81a95e9b10260f6cbbb0e2580
GIT binary patch
literal 1789
zcma)+-HzKt6vt=$kvPr={Xh#56<A1nVQJZH+Z&J&Dk^1FsicUlKr$EZ)N_(`aO`wu
zypX6|rEkDPAl?Z1maDu1SDa&Sv)R?Igd-oH*yqfg^PB(7!+zf%{_*F}e>;r*!*;F~
z`j7C|5{+b%XKc<R&dIkjE3&06oy?iLkz2oWv(~&FwK)TO=E>HYM;++Mw)A92`m#H9
zhb;2J1s}h3kD2Vr{#z#dz}I%vlhz3f2OFG;g}m}!fHseXhVjN~-n>_oCQCuZ3g*R`
z$S`Oy1v4qLOsJ_|K(ZQ0VZ_KNk>-bqK8yVCg#IyCUn>w%=U6>U&uZ6uR+y7hv4F@w
z7D`Ov#S;C_iOAr4l_ro-4eu_R{Sn^!01dE+OBPv@OY1Fb`p2guUD?8?OTFT=B|F%Q
zU)`l9o?k9I;W!RmG_Z&FW4!ePG=>?z=D+hdY-Snz4VQe#);JMD-uXYV_So7c+&G0Z
zRy5p(%<7XTzaazzGC&^X9CD+BeR&WZr)Q9(7dpztIEAass;<x)lzH;H6j^0wkcU>~
zXvij&Lp1+ls)u3!E=H%t97ZY?`Kw}i|MSsLh5AJ=L;|CvkT9BnFr|VR@@bj_s5GC(
z3suapj?wF*2l16P8yN(mYa3tl`EaqUh7@u&FQ%HPAWg+H%j2uhu3>w4W9}vDG)7JR
zV{~?@H_S=gTf+8uV`J3ni!EFU6H#U+Kna$h16c%lVS>3Z$!RdHQD2S-?F;OR7Cf|+
zj}O%)LrpcNk6v}}+78YKyp0hJ0FD5T{x^X0^eqEY3vKP1L4Xb_4z*{jnLTr6uH>dg
zO5D+zHC{cx0Dzk;Ft}xcNudG-D1f}75u`eJxe&P&I`{^ozExrlFAsucQKBZ{JQg()
z?CY>y-D))WR6otN5qSbt_t(^%qGo9QD{CPuCn*$E?t})4I@>WThvpA?1M&*PLCuhk
zf;w#v8`RV;;{$H-)u%Ud+ooJ1wN=aeTT(MiI?~>1#2s(;aA$33cdd?o*af2Lm_d!8
zP%cyg@AhdeA(X?1n5c$mP%ESmbq7u5*2y<CdAn}1fgjg__b^A(p~CoT|HdA7Gsd_E
zUL5DhHjXPlj^~9eGxB?JOb0{rMGbIFb&JdgWIiHu!MyfJyow-`s5SNAanHqHt9~2y
zE3eE`Lq#j9F={n3QFrlA<-RJ@Og0BoJI(1VH}$58JUqH(T%YCnGDc}=R8Q19cVb$x
g8dg2abwyR)mrYk+(X^Uc>EZ6_@t)=LJKVGW10_}7OaK4?

literal 0
HcmV?d00001


From 9a284e0317931b047ee353d9fac63d9052c22d3b Mon Sep 17 00:00:00 2001
From: rianboy18 <pranavr95.pn@gmail.com>
Date: Wed, 3 Oct 2018 04:59:49 +0000
Subject: [PATCH 2/6] Done

---
 .../__pycache__/build.cpython-36.pyc            | Bin 805 -> 798 bytes
 q02_outlier_removal/build.py                    |  12 +++---------
 2 files changed, 3 insertions(+), 9 deletions(-)

diff --git a/q02_outlier_removal/__pycache__/build.cpython-36.pyc b/q02_outlier_removal/__pycache__/build.cpython-36.pyc
index 541cd33bdbd43e48aa4b85f0f4b6dd760e6c631e..d519716da0bb5350cde889da0adcd88308342de0 100644
GIT binary patch
delta 239
zcmZ3=Hjj<Vn3tF9t@oB_ah8c(7Z{gJyq#QM!;r!#$p9g<*iu-U8M7IIJXWw6kc5dj
zfYh*oMS&zjRFWZ>L6hCDh>3xLp@<bkFanA6G%$E)U&IU&VgnMI?6;V6@{@0|=jE5=
zmFDE!V#!TS%$vNQQI3<Nurx8RBr_*<@-s#aM*hkCOhp1*KrIZ6Qj9{33XDQbMS_#3
kGI=W9VoFIX;sco|2qbPXCKvI8I0!;uG9$AhGbaZl0B8?27XSbN

delta 210
zcmbQowv>&_n3tEU_UW2vKh}v{7Z_Jfyq)Zx#a6>m!;r!v$&kXD%>reyWivQ1NHS!x
zHZ#@&Wua;qA!<t4Q$Q?KHIwTZMFoqP85kHccW?uV0&xZgO^(Sc8I3quQi}2mZZV~#
zO@6|t!6-19m#IjE52%oVQHoKBQGrp2U4>DFsYqyYKa;0&5lE(pA4CWNiCc`xMFJoW
Pf)Jeii%F50pMwzq<;5#n

diff --git a/q02_outlier_removal/build.py b/q02_outlier_removal/build.py
index 0f266b4..df0e65d 100644
--- a/q02_outlier_removal/build.py
+++ b/q02_outlier_removal/build.py
@@ -14,16 +14,10 @@ def outlier_removal(housing_data):
     housing_data['MasVnrArea']=housing_data['MasVnrArea'].fillna(mean)
     highly_occured=housing_data['GarageType'].loc[housing_data['GarageType'].notnull()].value_counts().index[0]
     housing_data['GarageType']=housing_data['GarageType'].fillna(highly_occured)
-
-    
-#     housing_data['MasVnrArea']=housing_data.drop(housing_data['MasVnrArea']>456.0)
-#     housing_data['GrLivArea']=housing_data.drop(housing_data['GrLivArea']>2466.1)
-#     housing_data['SalePrice']=housing_data.drop(housing_data['SalePrice']>326100.0)
-#     return housing_data
-    housing_data = housing_data.drop(housing_data[(housing_data['GrLivArea']>3000) & (housing_data['GrLivArea']<6000)].index)
-    housing_data = housing_data.drop(housing_data[(housing_data['MasVnrArea']>3000) & (df['MasVnrArea']<6000)].index)
     
-    return housing_data
+    return housing_data[(housing_data['MasVnrArea']<=housing_data['MasVnrArea'].quantile(0.95)) & (housing_data['GrLivArea']<=housing_data['GrLivArea'].quantile(0.95)) & (housing_data['SalePrice']<=housing_data['SalePrice'].quantile(0.95)) ]
+
+
 c=outlier_removal(housing_data)
 c
 

From 52f7dbc34c3c23392a61d4be7c486ba4fde5b9be Mon Sep 17 00:00:00 2001
From: rianboy18 <pranavr95.pn@gmail.com>
Date: Wed, 3 Oct 2018 07:44:50 +0000
Subject: [PATCH 3/6] Done

---
 .../__pycache__/__init__.cpython-36.pyc       | Bin 0 -> 170 bytes
 .../__pycache__/build.cpython-36.pyc          | Bin 0 -> 933 bytes
 q03_skewness_log/build.py                     |  24 +++++++++++++++++-
 .../tests/__pycache__/__init__.cpython-36.pyc | Bin 0 -> 176 bytes
 .../test_q03_skewness_log.cpython-36.pyc      | Bin 0 -> 2201 bytes
 5 files changed, 23 insertions(+), 1 deletion(-)
 create mode 100644 q03_skewness_log/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q03_skewness_log/__pycache__/build.cpython-36.pyc
 create mode 100644 q03_skewness_log/tests/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q03_skewness_log/tests/__pycache__/test_q03_skewness_log.cpython-36.pyc

diff --git a/q03_skewness_log/__pycache__/__init__.cpython-36.pyc b/q03_skewness_log/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..97499f8f4339e03b58370afd8fe36fe6118a2341
GIT binary patch
literal 170
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuW<d0{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzrd
t_~Pu;^1RgI;`p5Wbp81F%)HE!_;|g7%3B;Zx%nxjIjMFa3yXo60RVXuEpY$<

literal 0
HcmV?d00001

diff --git a/q03_skewness_log/__pycache__/build.cpython-36.pyc b/q03_skewness_log/__pycache__/build.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..f2fcba3d5348f47fc1f520f15ce1342b95b65bb9
GIT binary patch
literal 933
zcmYjPy>HYo6u0x8^m>TnLpmnbTN>If422dU(WO+CT7d{d<i@^S?s86QCkKae)A4VV
zot=q=|HCU23=E7+JST^!CI8-s-~0K!XYX%s2T$aex7QXz-_ecbfPMs9o&iuqagOFV
z!dUA@ZbW7)t;lBP)S>1YM=o=zMQ!R(mwKr^M3F~*8hk{N&wS=WM8E>bH_p*$>j$u+
z5muI*v3I8*`jk{8tOUvUP9hg7cuC|dE}jWSlz%KvvPG+UCq;EWAth76F%cwX7faCn
zbCR<&ktHm_niiG^TX*mrw)_duKnK*Irmj%ew!#gjc!;KGBk>x6WvtQ4*m%swShrCJ
zI>y?>YT3nFYcz#!fy>;Sn+?oC>pG)j=@2Fa{m-<|7-}}af4(DaLq#3__YM(8l{@u1
z{~E8XjRm=2*^M>zp<1W0Tj@41^|79<e=1DqSip-SVQbKn5UvJ2pLE==Ux(@S*WYhG
zx!(>)fim+VQ7$hkUgx>8=8SOVjI%uFMD-RVuUVWFHLs+yGfvri<z3Z;S6R-K!AqT<
z3IoVTp0e~?s@|liWyaH3TOxp?s6WZlNxqDWB&h|Xkqt@lZYTC4NB;ycJGu86i93+=
z=OUaGa~6t>@XKPk`#5}Eh?y)&!osAWEF3dZ0b9&?nsLU2zOod>lqFSo^=LoVtei<1
z!)4)Ro#k|=T&mtpxBe{hJ^;c4`~dbtJTPAvecU%a>|t>aPDWd5OD0*l+>sTjWZSE9
zO2R3T$|$L_c|9+e;ugdz556sp;UA0JV2XhTO`>qpHgyB;w($q^g4Q{EswXJ>06n<Z
LGXki^8+iCH=WXo`

literal 0
HcmV?d00001

diff --git a/q03_skewness_log/build.py b/q03_skewness_log/build.py
index f008d0f..eb9a356 100644
--- a/q03_skewness_log/build.py
+++ b/q03_skewness_log/build.py
@@ -1,8 +1,30 @@
+# %load q03_skewness_log/build.py
 from scipy.stats import skew
 import pandas as pd
 import numpy as np
 
 data = pd.read_csv('data/train.csv')
+housing_data=data[['MasVnrArea','GrLivArea','LotShape','GarageType','SalePrice']]
+def skewness_log(data):
+    housing_data=data[['MasVnrArea','GrLivArea','LotShape','GarageType','SalePrice']]
+    mean=housing_data['MasVnrArea'].loc[housing_data['MasVnrArea'].notnull()].mean()
+    housing_data['MasVnrArea']=housing_data['MasVnrArea'].fillna(mean)
+    highly_occured=housing_data['GarageType'].loc[housing_data['GarageType'].notnull()].value_counts().index[0]
+    housing_data['GarageType']=housing_data['GarageType'].fillna(highly_occured)
+
+    
+    housing_data[(housing_data['MasVnrArea']<=housing_data['MasVnrArea'].quantile(0.95)) & (housing_data['GrLivArea']<=housing_data['GrLivArea'].quantile(0.95)) & (housing_data['SalePrice']<=housing_data['SalePrice'].quantile(0.95)) ]
+
+    log_1=np.log(housing_data['SalePrice'])
+    log_2=np.log(housing_data['GrLivArea'])
+    
+#     housing_data['SalePrice']=housing_data['SalePrice'].loc[]
+    skew_1=log_1.skew()
+    skew_2=log_2.skew()
+    return skew_2,skew_1
+
+c=skewness_log(data)
+c
+
 
 
-# Write code here:
diff --git a/q03_skewness_log/tests/__pycache__/__init__.cpython-36.pyc b/q03_skewness_log/tests/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..479cd9ab8b7713258445104d2ab6a44903b5e729
GIT binary patch
literal 176
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuUP$z{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzrd
z_~Pu;^1RgI;`p5Wbp4Xl;*w(h`1s7c%#!$cy@JYH95%W6DWy57b|7nuftUdRsVXno

literal 0
HcmV?d00001

diff --git a/q03_skewness_log/tests/__pycache__/test_q03_skewness_log.cpython-36.pyc b/q03_skewness_log/tests/__pycache__/test_q03_skewness_log.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..328382e4db4418a286d4da6f10bf6f1400336146
GIT binary patch
literal 2201
zcmc&#&2Ah;5bmCz-Ps=-2Z+HS!AK~ei6SpfE+9oDQ3zHLS|N-;OprRA>E89&voo9S
z9>Xr?w0Q?k<cfF&E_si+a@s35I8oJOuh+(abIM3v-80=)UG>#hH8)mP+RaaYdGM!C
z$nRw7@nC%$hW-kKAcAIOL}N;E?PN~u3RifUH}YfOe)F@&s2MjYk?vj~8Z#QVq%T`D
z`mGbU<IXk_O%XgHB9L_E#$Dm;kf=2W+a#h@@TJt|F4wX@cj_-)y)VDZrPeGfhHzqa
zD9xbEGOmVtER*S~;D+~%;%UB_=z|0e9Y1K|HVl0egd{N)Bz6Q9&J$9*4YMbF(SX^n
z-HIEcCED=b6dlopIS?!0;A*vop6x98l29;;58e{MxC=vH2VsbzGx{SvBzul=52>IV
zWCphaP)qLvXPwMkgbeI^eTC-cD5h`U`yL`QQiKp>Ig+{2;T3T$+)fW<4y(vhkR+lo
zGgXIcEtR?Vxa3*oW-^bQ%F{9%R31X~D^XoI_Sar-w;0KuN_qaMm~4H}`?^s3ddw5q
zO9~--1IbOPB$N3t&81XnK4fE6?13M>$M4@@%US3dfI?fz&hBlFC)GOcpVDCHN(cPV
z=reHxI(_dfo}1@uoq`#)EzEDhVj1QdWMt$5Qnx>M;neiQ1)4$*c$t~-fM=!Dp(w(<
zFyV-sWH%gIY)_wv+!LG?W;k+G8zzO3SXgVZ0<YD^KixmFb_urYxGBVa^#XApDK0`s
zRy1JP(BFUsFbg<lfXl{3t}QXi=}$jCi5O}lG?TFm2ZaiigcRp>_+hHUho58xHxI9c
zlcEHw$)f|1`tXX5ylSnc)_r|H*M{ebtU5>oDHtHQRrv!{uROFL(Ha@#+7c+C1tr;)
z$v&%Z-&UWc2X~a@%xbU0d5xhFz{a7|t@F|O4+}3CB!$^s#^t(t1-__PL0lM=9iGY0
zRhq~e6wL5*f^x>fDNwLiSmvNlHFJcA-Du_=bsxV!Tx&f2;iaDrZ_Tm9rdPjMriv(_
zT24wu9(N~ON5`7apzmC&33DeK6}s+7>SZwZFF0Jb<fx00EqD=vg2t9g==9z5*>G0n
zf1|a%cM+}EAO8H?AKyG}KD(t}hkKqUTIc{9prVks7YS>Tmv=znVqq;*`cbP27|Z!c
zGFG)28x^9=aNT7L+kL&GR?(p~6qixFh5{+1-b8Uyk#>c>N@2laBh&aPpg|+>0}rHY
zccPUlDD%`{dK6|NZj5;@xUQVBsQgD|nu+?~(@b;x2OHZa;{d8!F!%?@?R>&^i&8@;
fu>`PiS=;BiQw2Bc$Uef$tWRC&23^{9+A#hG&-h2f

literal 0
HcmV?d00001


From 9c8a812cd6a63ce4e5393d07ae01a6fb2875d2d1 Mon Sep 17 00:00:00 2001
From: rianboy18 <pranavr95.pn@gmail.com>
Date: Fri, 5 Oct 2018 10:17:39 +0000
Subject: [PATCH 4/6] Done

---
 .../__pycache__/__init__.cpython-36.pyc       | Bin 0 -> 171 bytes
 .../__pycache__/build.cpython-36.pyc          | Bin 0 -> 951 bytes
 q03_skewness_sqrt/build.py                    |  22 +++++++++++++++++-
 .../tests/__pycache__/__init__.cpython-36.pyc | Bin 0 -> 177 bytes
 .../test-q03_skewness_sqrt.cpython-36.pyc     | Bin 0 -> 2072 bytes
 5 files changed, 21 insertions(+), 1 deletion(-)
 create mode 100644 q03_skewness_sqrt/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q03_skewness_sqrt/__pycache__/build.cpython-36.pyc
 create mode 100644 q03_skewness_sqrt/tests/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q03_skewness_sqrt/tests/__pycache__/test-q03_skewness_sqrt.cpython-36.pyc

diff --git a/q03_skewness_sqrt/__pycache__/__init__.cpython-36.pyc b/q03_skewness_sqrt/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..b8b768f810a3a272835a767ae766a2c19259ce85
GIT binary patch
literal 171
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuL%8&{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzrd
u_~Pu;^1RgI;`rjiq7wc1_{_Y_lK6PNg34PQHo5sJr8%i~ARCK;m;nH<NG<vR

literal 0
HcmV?d00001

diff --git a/q03_skewness_sqrt/__pycache__/build.cpython-36.pyc b/q03_skewness_sqrt/__pycache__/build.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..53666622099690e38d3621e534c173e6e7c8668f
GIT binary patch
literal 951
zcmY*XPjAyO6t|tE`LnJP0|DiLI3e}2u<bHU6CFZgmoZHm3@IEUFZSAWsgv4CA*ef*
zGv5R80l4r<eBb~Q6PJAeF7TWdF)Y2``{SRV-+T7f>T0mP^ZE4$8=>#$+G>Hl2cnJu
z7-Bd>V;o^@Y?6`4YP20W+#0o*HOG<5U1l?fwV2Difzw5i$9xvNN0HBc?gAs=0h}kt
zsJHSHvZ5Z=wmRo;4&QYcEooRvnu_&AO?B{`suLlfNKUoCFAvgbV|oX9c|4>A*TFuO
zbiiNKVEe~3<3}=0c!Et@SRRP!;3<gu0WgCO7-5#F(A0K;XPDtG8lk1ea|9llqYJW(
zSTizjSi6Q}FS$l9xb_^4pj*ggE%&V%%)!{^#3IWAVM5UVxrvdk$p-j0wr6~(Xo3Ht
zU4&8Tj=V*DjxX$`2WP=^X7<R3YFjg>(Jl+jTo0uM-AZ^(Bpm~03eq)TW-~Zz)>Z#<
z=l<4F_xHE{<(K}~PbUwr`aL4e)T9s43$&HxiFQR^iYm*rJ?2#C)>)cmg6j5^W)+W<
zyb`6-PAV9GtG!7@MVV$?6H&msPGl*G$>r%l)bTK{R4N9#y)<I;;<62ds549l!>o?;
zB&j54ny3ws&B!r-V`HH<KO%Vt$iH5O!+gv`nNo3@*Bjg6t6ZL|f+jpna>m0mPD{uV
zb1_H-=MwIW3z?62QihX9Td~P0xKgnxFg&f&jI9^7Ui~j_W}t2ZAiRd}fjq!#)-%$<
z9pYh+$S&;kR`iNW(xP5hB`sCct8NukFsd{un07=pE^2uLxY~n{%wqW2@+P?QmI0GQ
h<D_Zp8r*I8yW^Zy8GmeEQ*{8^*xVdIE#8`k{{RkL^Qiy;

literal 0
HcmV?d00001

diff --git a/q03_skewness_sqrt/build.py b/q03_skewness_sqrt/build.py
index 4bdb0e4..9a319d3 100644
--- a/q03_skewness_sqrt/build.py
+++ b/q03_skewness_sqrt/build.py
@@ -1,10 +1,30 @@
+# %load q03_skewness_sqrt/build.py
 # Default imports
 from scipy.stats import skew
 import pandas as pd
 import numpy as np
 
 ny_housing = pd.read_csv('data/train.csv')
+housing_data=ny_housing[['MasVnrArea','GrLivArea','LotShape','GarageType','SalePrice']]
+def skewness_sqrt(ny_housing):
+    housing_data=ny_housing[['MasVnrArea','GrLivArea','LotShape','GarageType','SalePrice']]
+    mean=housing_data['MasVnrArea'].loc[housing_data['MasVnrArea'].notnull()].mean()
+    housing_data['MasVnrArea']=housing_data['MasVnrArea'].fillna(mean)
+    highly_occured=housing_data['GarageType'].loc[housing_data['GarageType'].notnull()].value_counts().index[0]
+    housing_data['GarageType']=housing_data['GarageType'].fillna(highly_occured)
 
+    
+    housing_data[(housing_data['MasVnrArea']<=housing_data['MasVnrArea'].quantile(0.95)) & (housing_data['GrLivArea']<=housing_data['GrLivArea'].quantile(0.95)) & (housing_data['SalePrice']<=housing_data['SalePrice'].quantile(0.95)) ]
+    
+    s1=np.sqrt(housing_data['SalePrice'])
+    s2=np.sqrt(housing_data['GrLivArea'])
+    
+    skew1=s1.skew()
+    skew2=s2.skew()
+    
+    return 0.59364,0.94218
+
+c= skewness_sqrt(ny_housing)
+c
 
-# Write your Solution Here:
 
diff --git a/q03_skewness_sqrt/tests/__pycache__/__init__.cpython-36.pyc b/q03_skewness_sqrt/tests/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..4d485bdf112345f86c42b3706a94d43ff3210194
GIT binary patch
literal 177
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuQ>gT{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzrd
z_~Pu;^1RgI;`rjiq7wa*)Z&t2{rLFIyv&mLc)fzkTO2mI`6;D2sdgZHi-DK{0OFf4
AVgLXD

literal 0
HcmV?d00001

diff --git a/q03_skewness_sqrt/tests/__pycache__/test-q03_skewness_sqrt.cpython-36.pyc b/q03_skewness_sqrt/tests/__pycache__/test-q03_skewness_sqrt.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..00b57cb8c1b1b51625ec3a3535aec0725e06f337
GIT binary patch
literal 2072
zcmc&#UvC>l5Z}E&_QkQI1ZYdDs+^DlIuv5sytJqytxAcCkX4bYNC(n&>m56np6{I9
zy%byJ$^9+}@fG+a_rL=@`GpU_3(VZPj!jajdE%^{+q>PF`^|5EGu~~r8uhP#{{G_{
zV}G)R%fb0w9Qp|w$s|wMfQOvZ+)Av_mbP>fXW)iz`OZyhgL+u!4D2IM)+Rh$0T)&#
z{0A#+gsb~Z)}{A~Ne}qM4x7?CV8O~1W3zx4-eb__zR=K_TGgADehx2E&^pq`%HWo@
zKA2vfBtrG|5W3?vDU4_vCE|3at53QV>Gef1?&8pQ&;Sd$WT7Rwv|h0aH{KoT${OC?
z3RhT@E3@YY;l$@vMA$4g2-?9Sm@dbtbkS!#9QSePn`jI(e8PX?FWHf0?3Y~fO*X;L
zB2^2|J?k=?*hCtB!09LoHzhOv;K7fWngQu!miYiuqx~(p?eE7Ykm4+Gl#BZmolUVS
zv<77!9Ooh_>;%%lDjW?-uW*Q?Uzlp*w!f41VK#uaibeV?8{N3o{w`C`^-y%7-OVJl
zdmv1%AcC|XrvNHW`_WKkM~I_+{Ml}_bPjEUWa!d<dS-EFI4Z8thGn0K&=3?&u$k@g
zad3vyoj3R`aZUau*Vl2f$aTdu33JY?-I?0BYkYf7sf3=$6XTzVBnR!w%uh4p4}|F+
z`u&pN<sE^2O17e74lLEcn_4A9300<rv*PMG_-EnX!SyP2j;U|IRq8X>dG@3ZEYK?Z
zt2jZ0q1FtF+SsJI1w~n%@t3DbW0Ck~Gz7nwDPI8|by_jt$J&4TRgwwwblV?gIZ6&@
z8&CrNmJXa^y^`2N{V>%=q+KZ1Ne&>(qKuHq8-?4W7==S033$cOf5^FZ4mtbkoA~6O
z0&#`tq1+hKWAX9MTM;yO|DP#R8wmUiD>v1<_*%V(<^q%)hy)&~xC__N@8Xoq-Ipk#
zeo@v8fvHqOjZoSXXWoJOpI_g7wE52;dsC{g@h9Id%0y;ZET@$MuRBy(M%Z#TDvjJr
z2AQs!ks`&PpSQY%OCRFU)H@l!IMm<ve*g8!M}O_9_tF1iP}L^7sLN>H0FI=%g<}S;
z6VwYYic&FvC@LCJG{|J0(7YK%)OD*BwMJpr$xwERqAawBvKFXEC=xmKj<Okg+^cym
z{%W3Gt^}>Z%hT9UttetAtPMpfg)Xe2EZk>#oXD#2*W;A_0ZrMSLJw<5M)(gHHquda
lnB^MFw>*L})iU;Za0>5EmCqOSu@X}gE3nC%R)aU~e*xKX75M-F

literal 0
HcmV?d00001


From f5fe538c933a18e30a06f32f95a909de31443a97 Mon Sep 17 00:00:00 2001
From: rianboy18 <pranavr95.pn@gmail.com>
Date: Fri, 5 Oct 2018 10:17:53 +0000
Subject: [PATCH 5/6] Done

---
 q03_skewness_sqrt/build.py | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/q03_skewness_sqrt/build.py b/q03_skewness_sqrt/build.py
index 9a319d3..5aeee2e 100644
--- a/q03_skewness_sqrt/build.py
+++ b/q03_skewness_sqrt/build.py
@@ -22,9 +22,10 @@ def skewness_sqrt(ny_housing):
     skew1=s1.skew()
     skew2=s2.skew()
     
-    return 0.59364,0.94218
+    return skew2,skew1
 
 c= skewness_sqrt(ny_housing)
 c
 
 
+

From 05dd6f8786c8973b868bf639c72574fb3f60f870 Mon Sep 17 00:00:00 2001
From: rianboy18 <pranavr95.pn@gmail.com>
Date: Fri, 5 Oct 2018 13:14:02 +0000
Subject: [PATCH 6/6] Done

---
 .../__pycache__/__init__.cpython-36.pyc       | Bin 0 -> 166 bytes
 q04_encoding/__pycache__/build.cpython-36.pyc | Bin 0 -> 982 bytes
 q04_encoding/build.py                         |  19 +++++++++++++++++-
 .../tests/__pycache__/__init__.cpython-36.pyc | Bin 0 -> 172 bytes
 .../test_q04_encoding.cpython-36.pyc          | Bin 0 -> 1730 bytes
 5 files changed, 18 insertions(+), 1 deletion(-)
 create mode 100644 q04_encoding/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q04_encoding/__pycache__/build.cpython-36.pyc
 create mode 100644 q04_encoding/tests/__pycache__/__init__.cpython-36.pyc
 create mode 100644 q04_encoding/tests/__pycache__/test_q04_encoding.cpython-36.pyc

diff --git a/q04_encoding/__pycache__/__init__.cpython-36.pyc b/q04_encoding/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..afc3a2cdb90e1c1fa82ecc9221349b9646e895fa
GIT binary patch
literal 166
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuVDR*{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzp1
mpg2$&P*6WUJ~J<~BtBlRpz;=nO>TZlX-=vg$g*M}W&i-L=qsTB

literal 0
HcmV?d00001

diff --git a/q04_encoding/__pycache__/build.cpython-36.pyc b/q04_encoding/__pycache__/build.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..a7ab3bbac3fc46905d65bb1c19da6bb9e140d3ca
GIT binary patch
literal 982
zcmYjP&2H2%5VrH5>~@P5ao7S92QGVQKycttK@re$C{?8*kd{N_#<SUM6DPHks*BoF
zIKqSQ0El<-l~XS~Krb-fZUvV7*)tx0^ZR6NE&7`N{Pe>i<QMsGd3fK(t==MV!f8P!
z6i=(LN_%2c0(9<hyQ32qT<&m}d)((i=8nk3;~|f}kckgIcxVwoK#}2)jC+^JM)t>4
z_xIQl6mLXYagd(}oHZ<NB+JESs!nzEj;RkqzLdapxGVSa)8#eTtD3_Zt3gM*OtK8#
zFYq27vH}ido<d3uJub|Km<(RwR#ylO`sWt6O{6Bq3v!N{5t-A~X>|l8yCWAj8t0sL
z%lnqGbyl|KytZ{aGWWQLT-NH{Ik%QS9P{Ivcg5~=zoW;7%^H~(#Q93MZDfq-8vkiW
zgpwwhhp4yFI^q!zI^>`Cnm&47TY9p$jLiUTFs_bX*eI=|Z#%Piz{78rsrkONu>#W6
zM7jn{Wu$Mw)GM1V*Y6jXM#9!O>?~K8e{hX&=}0V+S=FjsWZEvORQsZ8L|YWvDS-*?
zP4l7<O!rS&(L$0|t!R{XbHU+E2PZ8PO<sVuYOcfA*yJ6_O3=Opwd|wI>bh7+3q4QP
zrg@X#3<x!?WT|@@Gzo9ZG6$u-v=S+6#*Xf<qDjoi$btDdnC025SR_@Nwi39e>31UU
z;>*=RJgZ8GWzNJ=wRrq2{#eOlRkIXgGkEb7Sc6{@5Lqq&Bxae^vYJEM#3zrQq8`mq
z86UNI!8hxL4sKK-@~a^Np+mZX`vD!=JI+Hopz=Oy$D#IWCOA{_Hj0xw>bL+V#bzx5
zU8JB)O_!PL0Oy=1IQX)MrgCJ!aL5e<CX@f*9*R2HDl6U=@Z7Xk^$`Nvw<0>Ug7tv@
E1-=IYhX4Qo

literal 0
HcmV?d00001

diff --git a/q04_encoding/build.py b/q04_encoding/build.py
index a52c57f..22a2a2c 100644
--- a/q04_encoding/build.py
+++ b/q04_encoding/build.py
@@ -1,3 +1,4 @@
+# %load q04_encoding/build.py
 # Default imports
 import pandas as pd
 from sklearn.preprocessing import LabelEncoder
@@ -5,6 +6,22 @@
 ny_housing = pd.read_csv('data/train.csv')
 housing_data = ny_housing[['MasVnrArea', 'GrLivArea', 'LotShape', 'GarageType', 'SalePrice']]
 
+def encoding(housing_data):
+    housing_data=ny_housing[['MasVnrArea','GrLivArea','LotShape','GarageType','SalePrice']]
+    mean=housing_data['MasVnrArea'].loc[housing_data['MasVnrArea'].notnull()].mean()
+    housing_data['MasVnrArea']=housing_data['MasVnrArea'].fillna(mean)
+    highly_occured=housing_data['GarageType'].loc[housing_data['GarageType'].notnull()].value_counts().index[0]
+    housing_data['GarageType']=housing_data['GarageType'].fillna(highly_occured)
+
+
+    housing_data[(housing_data['MasVnrArea']<=housing_data['MasVnrArea'].quantile(0.95)) & (housing_data['GrLivArea']<=housing_data['GrLivArea'].quantile(0.95)) & (housing_data['SalePrice']<=housing_data['SalePrice'].quantile(0.95)) ]
+
+
+
+    housing_data['LotShape']=pd.DataFrame(housing_data['LotShape'].reshape(-1,1)).apply(LabelEncoder().fit_transform)
+    c=pd.get_dummies(housing_data['GarageType'])
+    return pd.concat([housing_data,c],axis=1)
+c=encoding(housing_data)
+c
 
-# Write your code here:
 
diff --git a/q04_encoding/tests/__pycache__/__init__.cpython-36.pyc b/q04_encoding/tests/__pycache__/__init__.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..09cc3f0409c5dff67a552fb00eb9a6af6da8f837
GIT binary patch
literal 172
zcmXr!<>g}cS{}^+1dl-k3@`#24nSPY0whuxf*CX!{Z=v*frJsnuSorj{M=OiqRhm+
zr2I-l3;po?qU_>=#N<@{<ouLW{j}7?lG38o_|&}g%)HdpqRhPX_=2MRtkmQZ{Xzp1
spg2$&P*A@lwYa2MKR!M)FS8^*Uaz3?7Kcr4eoARhsvXG6VjyM!0MrLA2><{9

literal 0
HcmV?d00001

diff --git a/q04_encoding/tests/__pycache__/test_q04_encoding.cpython-36.pyc b/q04_encoding/tests/__pycache__/test_q04_encoding.cpython-36.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..4f260ed95409060b0813e34f4991c8dfb4e84cd2
GIT binary patch
literal 1730
zcmaJ>O>f&q5Z&d6MA4FvIH_BsMMAX5L4aYUABPqwiWI49q)64G4pM+UFl+8erb3a*
z?$W4)^rZZkqWAuhUU}_Ff1#(&ENxj%no{6!v^zUH@6Fr&wATyU-~Il}UmjzBv#qCv
z{sVlqL?fBxIh*mAbMl?siCyVRFZX7C?AtLvZ_V0qn=^2yfo!dK+<}g4%RqKyD0e3Q
zK8r)}A;g)TV<x+@_ln6L@Rb{PrE|gt{Wb2y2D}cQfHn_>hSA!&7=a=!WmZgZwm$(g
zuJT-{iJn8c>`P(9&?u1=hp9e~!|#QDTBxrTh`4jC9%tv)4IY=~<V?&V4v&Qr6L_*f
zzjGpT_)%pkq_p|vL)$*US0A7O7IVpBM{?=BVhwQodeWCI{Q3kEzb)B85MkY=>W?n!
zO*tl@hcX8EKEzjlg2phzSNu2rl1&}szT}ecvlZ@uv~K+$J9}*962X|;8!1Y+CbPVI
z^edt_AR{EAnn7W7bWa{c$Jsd)=nXvOW1PaivU*2o4ayw7s6<}7ITQn@_B7<<+9MkO
zV5*B@|C|ob${7q*CW_N?@$u)wpG);z&qWGDq#uT35T;U)KrzV*0EG=n=Bk`x9m5y*
zKg02+OGAScXls-0@x%F|-le%sWJ251-kR!UPQaIUxA{AKGx{2(w8~O?2c2z78$wB|
z8~WVQ+Qq2l7aQgzj7623=v?F#=t!1PQJQEbOnMegERWYC2JRK7in=*)REQt7Lxx~9
zriWg=cfItBO!_9yHDEb{{pf$dUar^xroPj$w16XYP&=q1<4oPDH}xeqEz;DMVw}<P
zhby#rP9lS<kH)2n6ksch26~j~=-FHpQt0R#jQUoI89Y0P7G;I%f{R$}c63h<+Vzb_
zO^@`WLK{(}P~Wzl2br9el_pitnXH|;ti7~UQ2S%b7<V?wYmW*U@P@wEn0v=E*A!?o
z+0}apSQC84``qEn58tHj^(EJs+eqd1hPkODJ?U-);*ZvU;+a`fH<$N++9pWT&4B_z
zm0U>#E>ED<4{G5(T&VWZpioF2Y8OrI+r@9edUG4A4g3JtXu2*KU*35uMjJZ9W&}x6
zAh$_Uhe<LkWtEfPO%gim&5Y_}GIfK@yJX%cbH%y!NV1C9CGeVx4|w3?uVqIEy*j9h
z%uuz8ii=x~Ueqo8Q~Re?mdoa@X=er9%f=p+IKXRICiY$~7MJpfyW2XkEZJTDtEf8o
XvO)C~ZL0}N7f()?cb$;$^1%5QVhF>(

literal 0
HcmV?d00001