6121123, 511 PM aboutblank Data Science Methodologies for 15 mine This couse focuses on the Foundational Methodology for Data Science by John Rollins, which was introduced in the previous video, However, its not the only ‘ethology shat you will encoute in data seienee, For example, in data ming, the Cross ladusty Process fo Data Mining (CRISP-DM) methodology i widely sed ‘What is CRISP-DM? ‘The CRISP-DM methodology is a process aimed at increasing the se of data mining over a wide variety of business applications and industries. The intent isto take case specific seenarios and general behavior 1 make them domain neutral. CRISP-DM is comprised of six steps with an ett that has to implement in ‘onder to havea reasonable chance of success. The sx steps are shown in the following diagram: Fig] CRISP-DM model, IBM Knowledge Cemer, CRISP-DM Help Overview ‘Business Understanding This stage isthe mos important because this is where the intention ofthe projec is ouined. Foundational Methodology and CCRISP-DM are aligned here. Ie equires communication and clarity. The difficulty here is shat stakeholders have diferent objetves, biases, and modalities ‘of relaing information, They don’t all ee the same things or inthe same manner Withou clear, concise, and complete perspestive of what te project, {goals are resources wil be needlessly expended Data Understanding Dats understanding relies on business understanding, Daa is cllecte st this stage ofthe process. The understanding of what the business was and needs will determine what data i collected, ftom what soutes, and by whet methods. CRISP-DM combines the stages of Data Requirements, Duta Collection, and Data Understanding from the Foundational Methodology outline Data Preparation Once the data hasbeen collected, i must be transformed into & useable subset ules it is determined that more dats is needed. Once 3 ataet is chosen, it mast then be checked for questionable, missing o ambiguous eases. Data Preparation is common to CRISP-DM and Foundational Methodology 4. Modeling Once prepared for use, the data must be expressed through whatever appropriate models, give meaningful insights, and hopefully new ‘knowledge. Tiss the purpose of data mining to create knowledge information that has meaning and uly. The use of models reveals pater and siructres within the data that provide insight ino the features of interest. Model are selected on «portion oF the data and adjstments are made it ‘necessity, Mode selection ian art and seienee, Both Foundational Methodology and CRISP-DM ave required for the subsequent sage Evaluation The selected model must be tested. Ths is usually done by having a pre-stlected test, se o rum the tained model on. This will allow you to sec the effectiveness ofthe mode! ona set i sees as new: Rests fom this are used to determine efficacy ofthe model and foreshadows its role i the next, fd fil stage 6. Deployment Inthe deployment step, she models used on new data outside of the scope ofthe datact and by new stakeholders. The new interactions at thie pase might revel the now varbles and needs fr the dataset end model. Thote now challenges could inate revision of ether business needs and actions, or the mode! and data, or both, CCRISP-DM is highly flexible and cyclical mode, Flexibility is requited at each step along with communication to keep the project ontrack, At any ofthe ix stages, it may be necessary to revisit aneavir stage and make changes. The key point of tis proces is that is eyclcal; therefore, even atthe finish you ate having another business understanding encounter lo discuss the viability ater deployment. The journey continues. For mote information on CRISP-DM, goto: BM Knowledge Center = CRISP-DM Help Overview ‘about blank 126121723, 511 PM aboutlank Author(s) Joseph Santareangeto Change log Date Version Changed by Change Description 2020-10-21 2.0 Malika Singla Migrated Lab to Markdown snd added to couse repo in GitLab © IBM Corporation 2023. All rights reserved. ‘about blank 2