Machine Learning Engineer Nanodegree

Capstone Project

Keyvan Tajbakhsh
August 12th, 2019

Customer Segmentation for Arvato Financial Services

This project relies on identifying key differentiators that divide customers into groups that can be targeted. Information such as a customers demographics (age, race, religion, gender, family size, ethnicity, income, education level), geography (where they live and work), psychographic (social class, lifestyle and personality characteristics) and behavioral (spending, consumption, usage and desired benefits) tendencies are taken into account when determining customer segmentation practices.

We will use unsupervised learning techniques to describe the relationship between the demographics of the company's existing customers and the general geographical population of Germany. The datasets provided need to be treated and prepared before implementing machine learning algorithms.

Our cluster analysis will be used to implement our supervised learning algorithm. In this context we will train and implement a supervised algorithm able to predict if a customer will respond positively to the mail-order campaign or not (binary classification problem). Then we will create a benchmark model to compare our final result and test the data.

Getting Started

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Prerequisites

NumPy - A fundamental package for scientific computing with Python.
Pandas - A library providing high-performance, easy-to-use data structures and data analysis tools.
ScikitLearn - Simple and efficient tools for data mining and data analysis
Matplotlib - Matplotlib is a Python 2D plotting library which produces publication quality figures in a variety of hardcopy formats and interactive environments across platforms
Pickle - The pickle module implements binary protocols for serializing and de-serializing a Python object structure.
Sea Born - Seaborn is a Python data visualization library based on matplotlib. It provides a high-level interface for drawing attractive and informative statistical graphics.
boto3 - Boto is the Amazon Web Services (AWS) SDK for Python. It enables Python developers to create, configure, and manage AWS services, such as EC2 and S3. Boto provides an easy to use, object-oriented API, as well as low-level access to AWS services.
SageMaker - SageMaker Python SDK is an open source library for training and deploying machine learning models on Amazon SageMaker.

You will also need to have software installed to run and execute a Jupyter Notebook

If you do not have Python installed yet, it is highly recommended that you install the Anaconda distribution of Python, which already has the above packages and more included.

Code

The project is divided into two parts. The code is provided in the 1_Customer_Segmentation_Report.ipynb and 2_Supervised_Learning_Model.ipynb notebook file. You will also be required to use aws SageMaker platform in the section Linear Learner to execute the code. This section is executed on Amazon SageMaker platform notebook. LinearLearner is a buitlin algorithm and we are only able to train and deploy this algorithm on Amazon SageMaker.

Run

In a terminal or command window, navigate to the top-level project directory Capstone_Project/ (that contains this README) and run one of the following commands:

ipython notebook 1_Customer_Segmentation_Report.ipynb

jupyter notebook 1_Customer_Segmentation_Report.ipynb

This will open the Jupyter Notebook software and project file in your browser.

Data

In this project datasets are provided by Udacity and limited to this project.However it is composed of two parts and four datasets described as follows:

1 - Customer Segmentation Report (unsupervised learning):

• Udacity_AZDIAS_052018.csv: Demographics data for the general population of Germany; 891 211 persons (rows) x 366 features (columns) • Udacity_CUSTOMERS_052018.csv: Demographics data for customers of a mail-order company; 191 652 persons (rows) x 369 features (columns)

The general population dataset (AZDIAS) will be used to create our unsupervised model (PCA and K-means). Then customers dataset will be mapped into the model in order to identify patterns and relation between customers groups.

2 - Supervised Learning Model:

• Udacity_MAILOUT_052018_TRAIN.csv: Demographics data for individuals who were targets of a marketing campaign; 42 982 persons (rows) x 367 (columns) • Udacity_MAILOUT_052018_TEST.csv: Demographics data for individuals who were targets of a marketing campaign; 42 833 persons (rows) x 366 (columns).

Before implementing our supervised model, the same steps are applied (part one) to the train dataset. After that we train our binary classifier to create a benchmark, then our target model. Different metrics will be used to evaluate our model.

Features

KBA05_DIESEL : share of cars with Diesel-engine in the microcell
KBA13_BJ_2009 : share of cars built in 2009 within the PLZ8
KBA05_ANTG4 : number of >10 family houses in the cell
KBA13_VW : share of VOLKSWAGEN within the PLZ8
KBA05_HERST4 : share of European manufacturer (e.g. Fiat, Peugeot, Rover,...)
KBA13_KW_110 : share of cars with an engine power between 91 and 110 KW - PLZ8
KBA13_SITZE_5 : number of cars with 5 seats in the PLZ8
KBA13_KW_30 : share of cars up to 30 KW engine power - PLZ8
D19_GESAMT_OFFLINE_DATUM : actuality of the last transaction with the complete file OFFLINE
KBA13_KMH_140_210 : share of cars with max speed between 140 and 210 km/h within the PLZ8
KBA13_KRSHERST_FORD_OPEL : share of FORD/Opel (referred to the county average) - PLZ8
KBA05_SEG2 : share of small and very small cars (Ford Fiesta, Ford Ka etc.) in the microcell
KBA05_ZUL2 : share of cars built between 1994 and 2000
ZABEOTYP : typification of energy consumers
ORTSGR_KLS9 : size of the community
PLZ8_ANTG1 : number of 1-2 family houses in the PLZ8
D19_VERSAND_ONLINE_QUOTE_12 : amount of online transactions within all transactions in the segment mail-order
KBA13_ANZAHL_PKW : number of cars in the PLZ8
KBA13_CCM_1500 : share of cars with 1400ccm to 1499ccm within the PLZ8
D19_BANKEN_ANZ_12 : transaction activity BANKS in the last 12 months
KBA05_MOTRAD : share of motorcycles per household
KBA13_KW_80 : share of cars with an engine power between 71 and 80 KW - PLZ8
KBA13_MOTOR : most common motor size within the PLZ8
GREEN_AVANTGARDE : Green avantgarde
KBA05_KRSKLEIN : share of small cars (referred to the county average)
KBA05_MAXBJ : most common age of the cars in the microcell
KBA13_KW_121 : share of cars with an engine power more than 120 KW - PLZ8
SEMIO_TRADV : affinity indicating in what way the person is traditional minded
KBA05_ANTG2 : number of 3-5 family houses in the cell
SEMIO_DOM : affinity indicating in what way the person is dominant minded
KBA13_HALTER_50 : share of car owners between 46 and 50 within the PLZ8
D19_VERSAND_ONLINE_DATUM : actuality of the last transaction for the segment mail-order ONLINE
FINANZ_SPARER : financial typology: money saver
KBA13_KMH_180 : share of cars with max speed between 110 km/h and 180km/h within the PLZ8
PLZ8_ANTG4 : number of >10 family houses in the PLZ8
KBA13_SEG_OBEREMITTELKLASSE : share of upper middle class cars and upper class cars (BMW5er, BMW7er etc.)
KBA05_SEG4 : share of middle class cars (Ford Mondeo etc.) in the microcell
KBA13_HALTER_65 : share of car owners between 61 and 65 within the PLZ8
ANZ_HAUSHALTE_AKTIV : number of households in the building
KBA13_KW_90 : share of cars with an engine power between 81 and 90 KW - PLZ8
SHOPPER_TYP : shopping typology
KBA05_KRSOBER : share of upper class cars (referred to the county average)
FINANZ_MINIMALIST : financial typology: low financial interest
GEBAEUDETYP : type of building (residential or commercial)
EWDICHTE : density of inhabitants per square kilometer
KBA13_KRSSEG_VAN : share of vans (referred to the county average) - PLZ8
KBA13_VORB_2 : share of cars with 2 preowner - PLZ8
LP_STATUS_GROB : social status rough
FINANZ_VORSORGER : financial typology: be prepared
PRAEGENDE_JUGENDJAHRE : dominating movement in the person's youth (avantgarde or mainstream)
KBA05_ALTER4 : share of cars owners elder than 61 years
KBA13_SITZE_4 : number of cars with less than 5 seats in the PLZ8
OST_WEST_KZ : flag indicating the former GDR/FRG
KBA05_AUTOQUOT : share of cars per household
KBA13_BJ_2004 : share of cars built before 2004 within the PLZ8
KBA13_KW_120 : share of cars with an engine power between 111 and 120 KW - PLZ8
KBA05_GBZ : number of buildings in the microcell
D19_TELKO_ONLINE_DATUM : actuality of the last transaction for the segment telecommunication ONLINE
KBA05_KRSAQUOT : share of cars per household (reffered to county average)
D19_BANKEN_DATUM : actuality of the last transaction for the segment banks TOTAL
KBA05_HERST5 : share of asian manufacturer (e.g. Toyota, Kia,...)
KBA05_MOD3 : share of Golf-class cars (in an AZ specific definition)
KBA05_SEG5 : share of upper middle class cars and upper class cars (BMW5er, BMW7er etc.)
KONSUMNAEHE : distance from a building to PoS (Point of Sale)
CAMEO_DEU_2015 : CAMEO classification 2015 - detailled classification
SEMIO_KRIT : affinity indicating in what way the person is critical minded
AGER_TYP : best-ager typology
KBA13_FIAT : share of FIAT within the PLZ8
HEALTH_TYP : health typology
ALTERSKATEGORIE_GROB : age classification through prename analysis
KBA13_HALTER_20 : share of car owners below 21 within the PLZ8
SEMIO_KULT : affinity indicating in what way the person is cultural minded
KBA13_NISSAN : share of NISSAN within the PLZ8
D19_BANKEN_OFFLINE_DATUM : actuality of the last transaction for the segment banks OFFLINE
KBA13_HALTER_60 : share of car owners between 56 and 60 within the PLZ8
FINANZ_UNAUFFAELLIGER : financial typology: unremarkable
KBA05_KRSHERST1 : share of Mercedes/BMW (reffered to the county average)
KBA05_MOD2 : share of middle class cars (in an AZ specific definition)
D19_VERSAND_ANZ_24 : transaction activity MAIL-ORDER in the last 24 months
KBA13_KW_0_60 : share of cars up to 60 KW engine power - PLZ8
KBA05_VORB0 : share of cars with no preowner
KBA13_BJ_2008 : share of cars built in 2008 within the PLZ8
KBA13_CCM_1200 : share of cars with 1000ccm to 1199ccm within the PLZ8
KBA13_KRSHERST_BMW_BENZ : share of BMW/Mercedes Benz (referred to the county average) - PLZ8
D19_GESAMT_ANZ_24 : transaction activity TOTAL POOL in the last 24 months
KBA05_SEG8 : share of roadster and convertables in the microcell
D19_VERSAND_OFFLINE_DATUM : actuality of the last transaction for the segment mail-order OFFLINE
SEMIO_KAEM : affinity indicating in what way the person is of a fightfull attitude
W_KEIT_KIND_HH : likelihood of a child present in this household
KBA13_MAZDA : share of MAZDA within the PLZ8
KBA05_ANTG3 : number of 6-10 family houses in the cell
KBA05_MOTOR : most common engine size in the microcell
ANZ_PERSONEN : number of adult persons in the household
KBA13_OPEL : share of OPEL within the PLZ8
KBA13_KMH_251 : share of cars with a greater max speed than 250 km/h within the PLZ8
KBA13_CCM_2501 : share of cars with more than 2500ccm within the PLZ8
KBA13_VORB_1 : share of cars with 1 preowner - PLZ8
KBA13_MERCEDES : share of MERCEDES within the PLZ8
KBA13_VORB_3 : share of cars with 3 or more preowner - PLZ8
ONLINE_AFFINITAET : online affinity
PLZ8_ANTG3 : number of 6-10 family houses in the PLZ8
D19_TELKO_ANZ_12 : transaction activity TELCO in the last 12 months
KBA05_SEG3 : share of lowe midclass cars (Ford Focus etc.) in the microcell
KBA05_ZUL1 : share of cars built before 1994
KBA13_SEG_UTILITIES : share of MUVs/SUVs within the PLZ8
KBA05_HERSTTEMP : development of the most common car manufacturers in the neighbourhood
KBA05_MAXVORB : most common preowner structure in the microcell
KBA05_ANTG1 : number of 1-2 family houses in the cell
KBA05_MAXAH : most common age of car owners in the microcell
KBA13_KMH_250 : share of cars with max speed between 210 and 250 km/h within the PLZ8
KBA13_SEG_MITTELKLASSE : share of middle class cars (Ford Mondeo etc.) in the PLZ8
KBA13_SEG_MINIVANS : share of minivans within the PLZ8
RELAT_AB : share of unemployed in relation to the county the community belongs to
ANREDE_KZ : gender
GFK_URLAUBERTYP : vacation habits
KBA05_MOD1 : share of upper class cars (in an AZ specific definition)
KBA13_CCM_3001 : share of cars with more than 3000ccm within the PLZ8
KBA05_KRSVAN : share of vans (referred to the county average)
KBA13_CCM_3000 : share of cars with 2500ccm to 2999ccm within the PLZ8
KBA13_PEUGEOT : share of PEUGEOT within the PLZ8
KBA13_TOYOTA : share of TOYOTA within the PLZ8
KBA13_HALTER_35 : share of car owners between 31 and 35 within the PLZ8
KBA13_BJ_1999 : share of cars built between 1995 and 1999 within the PLZ8
KBA13_CCM_2500 : share of cars with 2000ccm to 2499ccm within the PLZ8
KBA05_MAXHERST : most common car manufacturer in the microcell
KBA13_RENAULT : share of RENAULT within the PLZ8
KBA13_HALTER_40 : share of car owners between 36 and 40 within the PLZ8
D19_VERSI_ANZ_24 : transaction activity INSURANCE in the last 24 months
D19_VERSAND_ANZ_12 : transaction activity MAIL-ORDER in the last 12 months
KBA13_HALTER_45 : share of car owners between 41 and 45 within the PLZ8
KBA13_SEG_KLEINWAGEN : share of small and very small cars (Ford Fiesta, Ford Ka etc.) in the PLZ8
D19_BANKEN_ANZ_24 : transaction activity BANKS in the last 24 months
KBA05_SEG10 : share of more specific cars (Vans, convertables, all-terrains, MUVs etc.)
KBA13_HERST_FORD_OPEL : share of Ford & Opel/Vauxhall within the PLZ8
KKK : purchasing power
KBA05_KW1 : share of cars with less than 59 KW engine power
KBA05_MAXSEG : most common car segment in the microcell
SEMIO_VERT : affinity indicating in what way the person is dreamily
KBA05_MOD4 : share of small cars (in an AZ specific definition)
D19_VERSAND_DATUM : actuality of the last transaction for the segment mail-order TOTAL
BALLRAUM : distance to next urban centre
KBA13_BMW : share of BMW within the PLZ8
KBA13_SEG_GELAENDEWAGEN : share of allterrain within the PLZ8
LP_LEBENSPHASE_GROB : lifestage rough
KBA13_BJ_2006 : share of cars built between 2005 and 2006 within the PLZ8
KBA05_ZUL4 : share of cars built from 2003 on
SEMIO_PFLICHT : affinity indicating in what way the person is dutyfull traditional minded
KBA05_SEG7 : share of all-terrain vehicles and MUVs in the microcell
MIN_GEBAEUDEJAHR : year the building was first mentioned in our database
KBA05_ALTER1 : share of car owners less than 31 years old
LP_LEBENSPHASE_FEIN : lifestage fine
KBA05_BAUMAX : most common building-type within the cell
D19_VERSI_ANZ_12 : transaction activity INSURANCE in the last 12 months
KBA13_KW_60 : share of cars with an engine power between 51 and 60 KW - PLZ8
ANZ_HH_TITEL : number of academic title holder in building
KBA13_SEG_GROSSRAUMVANS : share of big sized vans within the PLZ8
KBA05_CCM3 : share of cars with 1800ccm to 2499 ccm
KBA13_ALTERHALTER_45 : share of car owners between 31 and 45 within the PLZ8
KBA13_HALTER_66 : share of car owners over 66 within the PLZ8
MOBI_REGIO : moving patterns
CAMEO_DEUG_2015 : CAMEO classification 2015 - Uppergroup
KBA13_ALTERHALTER_61 : share of car owners elder than 61 within the PLZ8
ANZ_TITEL : number of professional title holder in household
SEMIO_REL : affinity indicating in what way the person is religious
KBA13_CCM_1800 : share of cars with 1600ccm to 1799ccm within the PLZ8
KBA13_HALTER_25 : share of car owners between 21 and 25 within the PLZ8
KBA13_HERST_EUROPA : share of European cars within the PLZ8
D19_TELKO_OFFLINE_DATUM : actuality of the last transaction for the segment telecommunication OFFLINE
KBA13_CCM_0_1400 : share of cars with less than 1400ccm within the PLZ8
D19_GESAMT_ANZ_12 : transaction activity TOTAL POOL in the last 12 months
KBA13_AUDI : share of AUDI within the PLZ8
KBA13_KRSZUL_NEU : share of newbuilt cars (referred to the county average) - PLZ8
GEBAEUDETYP_RASTER : industrial areas
FINANZ_ANLEGER : financial typology: investor
KBA13_ALTERHALTER_60 : share of car owners between 46 and 60 within the PLZ8
KBA13_FAB_ASIEN : share of other Asian Manufacturers within the PLZ8
FINANZTYP : best descirbing financial type for the person
KBA05_HERST3 : share of Ford/Opel
KBA13_CCM_1600 : share of cars with 1500ccm to 1599ccm within the PLZ8
FINANZ_HAUSBAUER : financial typology: main focus is the own house
KBA13_CCM_1400 : share of cars with 1200ccm to 1399ccm within the PLZ8
KBA13_KW_61_120 : share of cars with an engine power between 61 and 120 KW - PLZ8
KBA13_SEG_VAN : share of vans within the PLZ8
D19_GESAMT_ONLINE_DATUM : actuality of the last transaction with the complete file ONLINE
D19_TELKO_DATUM : actuality of the last transaction for the segment telecommunication TOTAL
KBA13_KW_70 : share of cars with an engine power between 61 and 70 KW - PLZ8
SEMIO_MAT : affinity indicating in what way the person is material minded
KBA05_MOD8 : share of vans (in an AZ specific definition)
KBA05_CCM1 : share of cars with less than 1399ccm
D19_BANKEN_ONLINE_DATUM : actuality of the last transaction for the segment banks ONLINE
PLZ8_GBZ : number of buildings within the PLZ8
KBA05_KRSHERST3 : share of Ford/Opel (reffered to the county average)
KBA05_VORB1 : share of cars with one or two preowner
KBA05_VORB2 : share of cars with more than two preowner
KBA05_KRSHERST2 : share of Volkswagen (reffered to the county average)
KBA05_KRSZUL : share of newbuilt cars (referred to the county average)
KBA13_CCM_1000 : share of cars with less than 1000ccm within the PLZ8
KBA13_HERST_ASIEN : share of Asian Manufacturers within the PLZ8
KBA13_HERST_BMW_BENZ : share of BMW & Mercedes Benz within the PLZ8
KBA13_KMH_140 : share of cars with max speed between 110 km/h and 140km/h within the PLZ8
KBA13_AUTOQUOTE : share of cars per household within the PLZ8
KBA13_FAB_SONSTIGE : share of other Manufacturers within the PLZ8
KBA13_KRSHERST_AUDI_VW : share of Volkswagen (referred to the county average) - PLZ8
KBA13_VORB_0 : share of cars with no preowner - PLZ8
KBA13_KW_40 : share of cars with an engine power between 31 and 40 KW - PLZ8
KBA05_MODTEMP : development of the most common car segment in the neighbourhood
KBA05_SEG6 : share of upper class cars (BMW 7er etc.) in the microcell
KBA13_SEG_SPORTWAGEN : share of sportscars within the PLZ8
CJT_GESAMTTYP : customer journey typology
KBA13_KMH_0_140 : share of cars with max speed 140 km/h within the PLZ8
D19_BANKEN_ONLINE_QUOTE_12 : amount of online transactions within all transactions in the segment bank
KBA05_CCM4 : share of cars with more than 2499ccm
KBA05_SEG9 : share of vans in the microcell
VERS_TYP : insurance typology
KBA05_KW2 : share of cars with an engine power between 60 and 119 KW
TITEL_KZ : flag whether this person holds an academic title
KBA05_HERST1 : share of top German manufacturer (Mercedes, BMW)
KBA13_SEG_KLEINST : share of very small cars (Ford Ka etc.) in the PLZ8
KBA13_SEG_SONSTIGE : share of other cars within the PLZ8
KBA13_CCM_2000 : share of cars with 1800ccm to 1999ccm within the PLZ8
D19_GESAMT_ONLINE_QUOTE_12 : amount of online transactions within all transactions in the complete file
KBA05_CCM2 : share of cars with 1400ccm to 1799 ccm
KBA05_ZUL3 : share of cars built between 2001 and 2002
LP_FAMILIE_FEIN : familytyp fine
KBA13_SITZE_6 : number of cars with more than 5 seats in the PLZ8
SEMIO_RAT : affinity indicating in what way the person is of a rational mind
KBA05_HERST2 : share of Volkswagen-Cars (including Audi)
KBA13_HALTER_30 : share of car owners between 26 and 30 within the PLZ8
D19_GESAMT_DATUM : actuality of the last transaction with the complete file TOTAL
INNENSTADT : distance to the city centre
KBA13_KW_50 : share of cars with an engine power between 41 and 50 KW - PLZ8
KBA13_ALTERHALTER_30 : share of car owners below 31 within the PLZ8
KBA13_SEG_OBERKLASSE : share of upper class cars (BMW 7er etc.) in the PLZ8
LP_FAMILIE_GROB : familytyp rough
NATIONALITAET_KZ : nationaltity (scored by prename analysis)
SEMIO_ERL : affinity indicating in what way the person is eventful orientated
ALTER_HH : main age within the household
KBA05_ALTER3 : share of car owners inbetween 45 and 60 years of age
KBA05_KW3 : share of cars with an engine power of more than 119 KW
WOHNLAGE : residential-area
HH_EINKOMMEN_SCORE : estimated household net income
KBA13_KRSSEG_KLEIN : share of small cars (referred to the county average) - PLZ8
KBA13_SEG_KOMPAKTKLASSE : share of lowe midclass cars (Ford Focus etc.) in the PLZ8
KBA05_ANHANG : share of trailers in the microcell
KBA05_FRAU : share of female car owners
D19_KONSUMTYP : consumption type
KBA13_VORB_1_2 : share of cars with 1 or 2 preowner - PLZ8
WOHNDAUER_2008 : length of residence
KBA05_SEG1 : share of very small cars (Ford Ka etc.) in the microcell
REGIOTYP : neighbourhood
KBA13_SEG_MINIWAGEN : share of minicars within the PLZ8
PLZ8_BAUMAX : most common building-type within the PLZ8
RETOURTYP_BK_S : return type
KBA13_FORD : share of FORD within the PLZ8
KBA13_HALTER_55 : share of car owners between 51 and 55 within the PLZ8
KBA13_HERST_AUDI_VW : share of Volkswagen & Audi within the PLZ8
KBA13_KMH_110 : share of cars with max speed 110 km/h within the PLZ8
PLZ8_ANTG2 : number of 3-5 family houses in the PLZ8
KBA05_ALTER2 : share of car owners inbetween 31 and 45 years of age
D19_TELKO_ANZ_24 : transaction activity TELCO in the last 24 months
KBA13_KMH_211 : share of cars with a greater max speed than 210 km/h within the PLZ8
KBA13_KRSAQUOT : share of cars per household (referred to the county average) - PLZ8
LP_STATUS_FEIN : social status fine
KBA13_BJ_2000 : share of cars built between 2000 and 2003 within the PLZ8
SEMIO_LUST : affinity indicating in what way the person is sensual minded
KBA13_SEG_WOHNMOBILE : share of roadmobiles within the PLZ8
SEMIO_FAM : affinity indicating in what way the person is familiar minded
PLZ8_HHZ : number of households within the PLZ8
KBA13_KRSSEG_OBER : share of upper class cars (referred to the county average) - PLZ8
KBA13_HERST_SONST : share of other cars within the PLZ8
GEBURTSJAHR : year of birth
SEMIO_SOZ : affinity indicating in what way the person is social minded
CJT_TYP_3: not described
VHN: not described
D19_GARTEN: not described
D19_TECHNIK: not described
CJT_TYP_5: not described
D19_VERSICHERUNGEN: not described
D19_BEKLEIDUNG_REST: not described
MOBI_RASTER: not described
D19_GARTEN_RZ: not described
D19_KINDERARTIKEL: not described
D19_REISEN_RZ: not described
D19_BANKEN_LOKAL: not described
UMFELD_JUNG: not described
D19_BUCH_CD: not described
KONSUMZELLE: not described
D19_SAMMELARTIKEL_RZ: not described
D19_RATGEBER: not described
KBA13_ANTG3: not described
VK_ZG11: not described
KBA13_BAUMAX: not described
D19_HANDWERK: not described
VK_DHT4A: not described
CUSTOMER_GROUP: not described
D19_KK_KUNDENTYP: not described
AKT_DAT_KL: not described
BIP_FLAG: not described
ANZ_KINDER: not described
CJT_TYP_1: not described
SOHO_FLAG: not described
RT_SCHNAEPPCHEN: not described
D19_TECHNIK_RZ: not described
KBA13_ANTG4: not described
D19_SCHUHE_RZ: not described
KBA13_CCM_1400_2500: not described
D19_BILDUNG: not described
D19_REISEN: not described
D19_BIO_OEKO: not described
HH_DELTA_FLAG: not described
CJT_KATALOGNUTZER: not described
ALTER_KIND2: not described
D19_RATGEBER_RZ: not described
D19_LETZTER_KAUF_BRANCHE: not described
D19_KONSUMTYP_MAX: not described
D19_FREIZEIT: not described
CAMEO_DEUINTL_2015: not described
KOMBIALTER: not described
D19_TELKO_REST: not described
D19_WEIN_FEINKOST: not described
D19_SOZIALES: not described
D19_BANKEN_REST: not described
VK_DISTANZ: not described
D19_KINDERARTIKEL_RZ: not described
D19_VOLLSORTIMENT_RZ: not described
D19_TIERARTIKEL: not described
CJT_TYP_2: not described
D19_KOSMETIK_RZ: not described
D19_FREIZEIT_RZ: not described
CAMEO_INTL_2015: not described
DSL_FLAG: not described
D19_LEBENSMITTEL_RZ: not described
D19_BANKEN_DIREKT: not described
D19_ENERGIE_RZ: not described
ALTER_KIND4: not described
KBA13_GBZ: not described
UNGLEICHENN_FLAG: not described
CJT_TYP_6: not described
D19_BANKEN_GROSS_RZ: not described
D19_VERSAND_REST: not described
D19_HAUS_DEKO_RZ: not described
D19_VERSI_OFFLINE_DATUM: not described
D19_KOSMETIK: not described
D19_LEBENSMITTEL: not described
D19_VERSI_ONLINE_QUOTE_12: not described
KBA13_KMH_210: not described
SOHO_KZ: not described
D19_TELKO_REST_RZ: not described
GEMEINDETYP: not described
D19_DIGIT_SERV: not described
D19_TELKO_ONLINE_QUOTE_12: not described
D19_LOTTO: not described
RT_UEBERGROESSE: not described
D19_VERSICHERUNGEN_RZ: not described
D19_HANDWERK_RZ: not described
D19_BANKEN_REST_RZ: not described
EXTSEL992: not described
D19_BEKLEIDUNG_GEH_RZ: not described
RT_KEIN_ANREIZ: not described
VHA: not described
KBA13_CCM_1401_2500: not described
KK_KUNDENTYP: not described
KBA13_ANTG2: not described
D19_BILDUNG_RZ: not described
D19_BEKLEIDUNG_GEH: not described
D19_SCHUHE: not described
D19_BUCH_RZ: not described
STRUKTURTYP: not described
ALTER_KIND1: not described
D19_VOLLSORTIMENT: not described
ALTER_KIND3: not described
UMFELD_ALT: not described
D19_LOTTO_RZ: not described
VERDICHTUNGSRAUM: not described
WACHSTUMSGEBIET_NB: not described
FIRMENDICHTE: not described
KBA13_ANTG1: not described
D19_NAHRUNGSERGAENZUNG: not described
D19_HAUS_DEKO: not described
HAUSHALTSSTRUKTUR: not described
D19_NAHRUNGSERGAENZUNG_RZ: not described
D19_VERSI_ONLINE_DATUM: not described
EINGEZOGENAM_HH_JAHR: not described
ONLINE_PURCHASE: not described
PRODUCT_GROUP: not described
ANZ_STATISTISCHE_HAUSHALTE: not described
D19_TIERARTIKEL_RZ: not described
EINGEFUEGT_AM: not described
D19_BEKLEIDUNG_REST_RZ: not described
D19_TELKO_MOBILE: not described
D19_SONSTIGE: not described
CJT_TYP_4: not described
D19_DIGIT_SERV_RZ: not described
D19_BANKEN_LOKAL_RZ: not described
ALTERSKATEGORIE_FEIN: not described
D19_DROGERIEARTIKEL: not described
KBA13_HHZ: not described
D19_WEIN_FEINKOST_RZ: not described
GEOSCORE_KLS7: not described
D19_BANKEN_DIREKT_RZ: not described
D19_BANKEN_GROSS: not described
D19_SONSTIGE_RZ: not described
D19_TELKO_MOBILE_RZ: not described
D19_SAMMELARTIKEL: not described
ARBEIT: not described
D19_ENERGIE: not described
D19_VERSAND_REST_RZ: not described
D19_DROGERIEARTIKEL_RZ: not described
D19_VERSI_DATUM: not described
D19_BIO_OEKO_RZ: not described

Target Variable

Recall: The recall is the ratio tp / (tp + fn) where tp is the number of true positives and fn the number of false negatives. The recall is intuitively the ability of the classifier to find all the positive samples.
Precision: The precision is the ratio tp / (tp + fp) where tp is the number of true positives and fp the number of false positives. The precision is intuitively the ability of the classifier not to label as positive a sample that is negative.
F1: The F1 score can be interpreted as a weighted average of the precision and recall, where an F1 score reaches its best value at 1 and worst score at 0. The relative contribution of precision and recall to the F1 score are equal. The formula for the F1 score is: F1 = 2 * (precision * recall) / (precision + recall)
Accuracy: Accuracy classification score.

keyvantaj/Capstone_Project