class: center, middle, title-slide # GEOG5927 Predictive Analytics: Closing ### Roger Beecham ### 25 Apr 2022 --- ## Module Schedule <table> <thead> <tr> <th style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 18px;"> session </th> <th style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 18px;"> wc </th> <th style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 18px;"> academic </th> <th style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 18px;"> lecture </th> <th style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 18px;"> deadline </th> </tr> </thead> <tbody> <tr> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 1 </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 28 Feb </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> RB </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> Simulating behaviour </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> </td> </tr> <tr> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 2 </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 7 Mar </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> RB </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> Targeted marketing </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> </td> </tr> <tr> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 3 </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 14 Mar </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> RO </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> Machine learning </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> </td> </tr> <tr> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 4 </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> 21 Mar </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> JG/NM </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> Agent-based models </td> <td style="text-align:left;color: #616161 !important;background-color: #ffffff !important;font-size: 20px;"> Ass 1 </td> </tr> <tr> <td style="text-align:left;background-color: #ffffff !important;font-size: 20px;"> 5 </td> <td style="text-align:left;background-color: #ffffff !important;font-size: 20px;"> 25 Apr </td> <td style="text-align:left;background-color: #ffffff !important;font-size: 20px;"> JG/NM </td> <td style="text-align:left;background-color: #ffffff !important;font-size: 20px;"> Close </td> <td style="text-align:left;background-color: #ffffff !important;font-size: 20px;"> Ass 2 </td> </tr> </tbody> </table> ??? * Give a quick recap/reflection over the module * Assignment 2 and presentations more generally * Module survey/evaluation --- ## Module content and philosophy * Spatial modelling to simulate and predict consumer behaviour + Exploratory analysis + Microsimulation + Agent-based modelling <br> * Research case studies to evaluate modelling techniques in practice + Practical sessions + Individual data science report + Group data science presentation ??? * Introduced some key theory and techniques in modern data analysis. * Both technical and applied aspects. * Techniques in which Geog dept specialises * Doing so with datasets and domains relevant to marketing science --- ## Outcomes By the end of this module you should be able to: 1. **Explain** and **critically evaluate** the role of spatial analytics in simulating and predicting consumer behaviours 2. **Apply** geocomputational modelling and simulation techniques on real data sets 3. **Devise** and **employ** spatial modelling tools to address business problems, presenting and justifying recommendations in an appropriate context -- `-----------` microsimulation | random forests | abms ??? 1 : theory 2 : application 3 : both --- ## Spatial microsimulation .pull-left[.right[ `Survey data` .small-font[ individual-level and rich in detail <br> small sample and may be biased ]] ] <img src = "img/icon_survey.jpg", width = 20%, style = "position:absolute; top: 28%; left: 52%;"></img> <br><br><br><br><br><br><br> .pull-right[.left[ `Census data` .small-font[ high-level and low in detail <br> population-level and complete ]] ] <img src = "img/icon_census.jpeg", width = 20%, style = "position:absolute; top: 65%; left: 28%;"></img> ??? Many situations where interested in knowing population that lives in an area: interests, preferences, spending patterns Are population-level datasets that exist.: Census is amazing for counting people according to high-level characteristics... But we only have a limited set of attribute information. Lots missed off from it – commercial setting – interests and preferences. Instead -- rely on comparatively small sample survey data for studying interests and preferences. Spatial microsimulation allows us to match rich individual-level data to a population we know less about. --- ## Random forests <img src = "img/ml_supervised.png", width = 40%, style = "position:absolute; top: 28%; left: 8%;"></img> <img src = "img/ml_rf.png", width = 40%, style = "position:absolute; top: 28%; left: 45%;"></img> --- ## Agent-based models <img src = "img/abm.png", width = 65%, style = "position:absolute; top: 25%; left: 8%;"></img> --- ## Fundamentals of (modern) data analysis -- * Data analysis is + A careful thinking about evidence (data) in the context of a research problem -- * Data analysis involves + **Defining** your problem + **Identifying** relevant data + **Winnowing** on data | techniques | problem-space + **Evaluating** implied patterns under uncertainty ??? these are the sorts of things that you will be doing in dissertations --- ## Fundamentals of (modern) data analysis <br> .small-font[**Deviation** from **Expectation**] <img src = "img/correl_heer.jpeg", width = 60%, style = "position:absolute; top: 35%; left: 5%;"></img> <br><br><br><br><br><br><br><br> .xtiny-font[Correll & Heer (2017) Surprise! Bayesian Weighting for De-Biasing Thematic Maps, IEEE TVCG] --- ## Fundamentals of (modern) data analysis <img src = "img/props.png", width = 20%, style = "position:absolute; top: 22%; left: 5%;"></img> -- <img src = "img/props_control.png", width = 10.75%, style = "position:absolute; top: 22%; left: 27%;"></img> -- <img src = "img/props_net.png", width = 10.75%, style = "position:absolute; top: 22%; left: 40%;"></img> <img src = "img/props_label.png", width = 7.5%, style = "position:absolute; top: 22%; left: 55%;"></img> --- ## Fundamentals of (modern) data analysis <img src="img/bikeshare.png", width = 60%, style = "position:absolute; top: 35%; left: 8%;"> </img> --- ## Fundamentals of (modern) data analysis <img src="img/bike_all.png", width = 60%, style = "position:absolute; top: 18%; left: 8%;"> </img> --- ## Fundamentals of (modern) data analysis <img src="img/bike_mf.png", width = 60%, style = "position:absolute; top: 18%; left: 8%;"> </img> --- ## Fundamentals of (modern) data analysis <img src="img/bike_legend.png", width = 60%, style = "position:absolute; top: 18%; left: 8%;"> </img> --- ## Fundamentals of (modern) data analysis <img src="img/bike6.jpeg", width = 60%, style = "position:absolute; top: 18%; left: 8%;"> </img> --- ## Assignment #2 -- ### Guidelines for effective presentations (slides) --- ## Assignment #2 ### Guideline 1: Avoid noise -- .small-font[ * Background colours * Logos * Overly small font * Too much text * Unnecessary transitions ] --- ## Assignment #2 ### Guideline 1: Avoid noise -- <img src = "img/excel_default.jpeg", width = 22%, style = "position:absolute; top: 35%; left: 8%; "></img> -- <img src = "img/remove_shadow.jpeg", width = 22%, style = "position:absolute; top: 35%; left: 33%; "></img> -- <img src = "img/emphasise_data.jpeg", width = 22%, style = "position:absolute; top: 35%; left: 58%; "></img> -- <img src = "img/design_purpose.jpeg", width = 22%, style = "position:absolute; top: 68%; left: 8%; "></img> -- <img src = "img/emphasise_patterns.jpeg", width = 22%, style = "position:absolute; top: 68%; left: 33%; "></img> --- ## Assignment #2 ### Guideline 1: Avoid noise <img src = "img/excel_default.jpeg", width = 30%, style = "position:absolute; top: 40%; left: 15%; "></img> <img src = "img/emphasise_patterns.jpeg", width = 30%, style = "position:absolute; top: 40%; left: 48%; "></img> --- ## Assignment #2 ### Guideline 2: Refine <br> > .small-font[ *With each slide,* <br> *convey one message (only)*] ??? Very easy to load slides with content. Try to be judicious with the message. --- ## Assignment #2 ### Guideline 3: Reduce <br> > .small-font[ *Be concise,* <br> *both verbally and visually*] ??? Say what you want to say with the least number of words. --- ## Assignment #2 ### Guideline 4: Compliment <br> > .small-font[ *Slides should display things that* <br> *can’t be easily spoken*] ??? Say what you want to say with the least number of words. --- ## Assignment #2 ### Guideline 5: Layout -- <img src = "img/layout1.jpeg", width = 60%, style = "position:absolute; top: 35%; left: 8%; "></img> -- <img src = "img/layout2.jpeg", width = 60%, style = "position:absolute; top: 85%; left: 8%; "></img> ??? Do think about ordering your slides meaningfully : layout is powerful. --- ## Assignment #2 ### Guideline 5: Layout -- <img src = "img/layout3.jpeg", width = 65%, style = "position:absolute; top: 50%; left: 8%; "></img> --- ## Assignment #2 ### Guideline 5: Layout <img src = "img/layout4.jpeg", width = 65%, style = "position:absolute; top: 50%; left: 8%; "></img> --- ## Assignment #2 ### Guideline 5: Layout - order <br> .small-font[ > *We expect things to be displayed in sequence.*] -- .small-font[ > *If we wish to imply a sequence, arrange things in that sequence.*] -- .small-font[ > *This can be particularly useful when ‘telling a story’ in a presentation.* ] ??? We expect a story, a sequence We can imply that sequence in our presentations by ordering slides and charts Doing so allows us to tell data stories. --- ## Assignment #2 ### Jean-Luc Doumont <div class="embed-responsive embed-responsive-16by9"> <iframe width="500" height="350" class="embed-responsive-item" src="https://www.youtube.com/embed/meBXuTIPJQk" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe> </div> <!-- <img src = "img/jean_luc.jpg", width = 40%, style = "position:absolute; top: 33%; left: 8%; "></img> --> --- <img src="img/eval.png", width = 75%, style = "position:absolute; top: 5%; left: 6%;"> </img> <br><br><br><br><br><br><br><br><br><br><br><br><br><br><br> `https://leeds.bluera.com/leeds/` ??? * opened today * automatic e-mail reminders * useful for updating content : initially this module used peculiar software drag and drop (no code), moved to R in response to feedback * do try to be specific / direct / focussed with comments --- and think carefully about expectations of you as MSc students