Modeling Specialties
Mixed models
Machine learning
High dimensional data
Code optimization
Text analysis
Analysis of complex sample surveys
GIS, spatial, remote-sensing
and more!
@@ -509,7 +509,7 @@
Workshops
-
Hands on programming
Data Science Skill Series
+
Hands-on programming
Data Science Skill Series
diff --git a/cscar_light.Rmd b/cscar_light.Rmd
new file mode 100644
index 0000000..6112f61
--- /dev/null
+++ b/cscar_light.Rmd
@@ -0,0 +1,322 @@
+---
+title: "
"
+
+author: "cscar.research.umich.edu"
+output:
+ revealjs::revealjs_presentation:
+ theme: white
+ transition: slide
+ incremental: true
+ center: true
+ css: slides.css
+---
+
+```{r setup, include=FALSE}
+knitr::opts_chunk$set(echo = TRUE)
+```
+
+## Data Science Collaboration
+
+
+
+
+
+
+
+## Statistical Modeling {data-background='img/priorLikePosterior.png' data-state='opac50'}
+
+
++
++
+
+ +
+ +$$y \sim \mathscr{N}(X\beta, \sigma^2)$$ +
+ + +## Modeling Specialties {data-background='img/ml2.png' data-state='opac25'} + +
+Mixed models
+Machine learning
+High dimensional data
+Code optimization
+Text analysis
+Analysis of complex sample surveys
+GIS, spatial, remote-sensing
+and more!
+
+R
+Python
+Matlab
+Stan
+Stata
+Go
+and more!
+
and
Computational Programming + +
+
++ +
+
+
+
+
+## Other Assistance {data-background='img/plotly_scatter.png' data-state='opac75'}
+
+
+
+Visualization
+Web scraping
+Data I/O
+Concept formation
+Reproducible research
+Best practices
+
+For common data and analysis problems, just stop by!
+
+Send a question to the help list!
+
+ +
+stats-consulting@umich.edu ds-consulting@umich.edu hpc-consulting@umich.edu +
+ +## Workshops {data-background='img/code_white2.png' data-state='opac50'} + + +
+
Hands-on programming
+Data Science Skill Series
+ +
+[cscar.research.umich.edu/events](cscar.research.umich.edu/events) +
+ +## {data-transition="zoom" data-autoslide="3000"} + + +
+
+
+## {data-transition="zoom" data-autoslide="3000" }
+
+
+
+## {data-transition="zoom" data-autoslide="2000"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="2000" }
+
+
+
+## {data-transition="zoom" data-autoslide="1000"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="1000" }
+
+
+
+## {data-transition="zoom" data-autoslide="500"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="500" }
+
+
+
+
+## {data-transition="zoom" data-autoslide="250"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="250" }
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="125" }
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="125" }
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125" }
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125" }
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125" }
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+## {data-transition="zoom" data-autoslide="125" }
+
+
+
+## {data-transition="zoom" data-autoslide="125"}
+
+
+
+
+
+## {data-transition="zoom" data-background="img/sunshine.jpg" data-autoslide="3000"}
+
+
+
+Data Science Serenity
+
+
+
+## Join Us Now! {data-transition="zoom" data-autoslide="3000" }
+
+Can you do any less?!
+ + + +## {data-background="img/cscar84.png" data-state='opac75'} \ No newline at end of file diff --git a/cscar_light.html b/cscar_light.html new file mode 100644 index 0000000..35ea849 --- /dev/null +++ b/cscar_light.html @@ -0,0 +1,721 @@ + + + + + + +
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/img/code_white.png b/img/code_white.png
new file mode 100644
index 0000000..87cfff4
Binary files /dev/null and b/img/code_white.png differ
diff --git a/img/code_white2.png b/img/code_white2.png
new file mode 100644
index 0000000..f16ba19
Binary files /dev/null and b/img/code_white2.png differ
diff --git a/misc_code_for_display.R b/misc_code_for_display.R
new file mode 100644
index 0000000..136047a
--- /dev/null
+++ b/misc_code_for_display.R
@@ -0,0 +1,266 @@
+# Functions ---------------------------------------------------------------
+
+
+# ints and Times
+dataGen <- function(nclusters,
+ nwithin,
+ corr = 0,
+ balanced = T)
+{
+ # setup
+ nclus = nclusters # number of groups
+ clus = factor(rep(1:nclus, each=nwithin)) # cluster variable
+ n = length(clus) # total n
+
+ # parameters
+ sigma = 1 # residual sd
+ psi = matrix(c(1,corr,corr,1), 2, 2) # re covar
+ gamma_ = MASS::mvrnorm(nclus, mu=c(0,0), Sigma=psi, empirical=TRUE) # random effects
+ e = rnorm(n, mean=0, sd=sigma) # residual error
+ intercept = 3 # fixed effects
+ b1 = .75
+
+ # data
+ x = rep(1:nwithin-1, times=nclus) # covariate
+ y = intercept+gamma_[clus,1] + (b1+gamma_[clus,2])*x + e # target
+ d = data.frame(time=x, y, clus=clus)
+}
+
+# nlme control msMaxIter, msMaxEval, abs.tol set to lavaan defaults
+
+
+runNLME <- function(data)
+{
+ nlmemod = lme(y ~ time, data=data,
+ random = ~time|clus,
+ weights=varIdent(form = ~1|time),
+ method='ML',
+ control = list(
+ maxIter = 1000,
+ msMaxIter = 10000,
+ msMaxEval = 20000,
+ returnObject = T,
+ abs.tol = 2.220446e-15
+ ))
+ varRes = coef(nlmemod$modelStruct$varStruct, unconstrained =FALSE,allCoef=T)*nlmemod$sigma
+ varRE = as.numeric(VarCorr(nlmemod)[1:2,1])
+ corRE = as.numeric(VarCorr(nlmemod)['time','Corr'])
+ fixed = fixef(nlmemod)
+ list(varRes=varRes, varRE=varRE, corRE=corRE, fixed=fixed)
+}
+
+runNLME_REML <- function(data) {
+ nlmemod = lme(y ~ time, data=data,
+ random = ~time|clus,
+ weights=varIdent(form = ~1|time),
+ method='REML',
+ control = list(
+ maxIter = 1000,
+ msMaxIter = 10000,
+ msMaxEval = 20000,
+ returnObject = T,
+ abs.tol = 2.220446e-15
+ ))
+ varRes = coef(nlmemod$modelStruct$varStruct, unconstrained =FALSE,allCoef=T)*nlmemod$sigma
+ varRE = as.numeric(VarCorr(nlmemod)[1:2,1])
+ corRE = as.numeric(VarCorr(nlmemod)['time','Corr'])
+ fixed = fixef(nlmemod)
+ list(varRes=varRes, varRE=varRE, corRE=corRE, fixed=fixed)
+}
+
+runGAM <- function(data) {
+ gammod = gam(y ~ time + s(clus, bs='re') + s(time, clus, bs='re'), data=data,
+ method='REML')
+ varRes = gammod$sig2
+ varRE = gam.vcomp(gammod)[1:2,1]^2
+ # corRE = as.numeric(VarCorr(nlmemod)['time','Corr'])
+ fixed = coef(gammod)[1:2]
+ list(varRes=varRes, varRE=varRE, corRE=NULL, fixed=fixed)
+}
+
+runBayes_brm = function(data, ...){
+ # variance heterogeneity coming in next release
+ bayesmod = brm(y ~ time + (1 + time|clus), data = data,
+ iter = 6000, warmup=1000, thin=20,
+ prior = set_prior("normal(0,10)", class = "b"),
+ ...)
+ vc = VarCorr(bayesmod)
+ varRes = VarCorr(bayesmod)[[2]][['cov']][['mean']]
+ varRE = diag(VarCorr(bayesmod)[[1]][['cov']][['mean']])
+ corRE = VarCorr(bayesmod)[[1]][['cor']][['mean']][2]
+ fixed = fixef(bayesmod)[,'mean']
+ list(varRes=varRes, varRE=varRE, corRE=corRE, fixed=fixed)
+}
+
+runBayes_rstan = function(compcode, data, ...){
+ standat = brms::make_standata(y ~ time + (1 + time|clus), data=data)
+ bayesmod = sampling(sm, data = standat,
+ iter = 6000, warmup=1000, thin=20,...)
+ varRes = get_posterior_mean(bayesmod, 'sigma')[,'mean-all chains']
+ varRE = get_posterior_mean(bayesmod, 'sd_1')[,'mean-all chains']
+ corRE = get_posterior_mean(bayesmod, 'cor_1')[,'mean-all chains']
+ fixed = get_posterior_mean(bayesmod, pars=c('b_Intercept', 'b'))[,'mean-all chains']
+ list(varRes=varRes, varRE=varRE, corRE=corRE, fixed=fixed)
+}
+
+
+runGC <- function(data) {
+ ntime = unique(data$time)
+ data$time = factor(data$time)
+ dataWide = tidyr::spread(data, time, y)
+ colnames(dataWide)[-1] = paste0('y', colnames(dataWide)[-1])
+
+ IModel = paste0('I =~ 1*y0 ', paste0('+ 1*', colnames(dataWide)[-c(1:2)], collapse=''), '\n')
+ SModel = paste0('S =~ 0*y0 ', paste0('+ ', 1:dplyr::last(ntime), '*', colnames(dataWide)[-c(1:2)], collapse=''), '\n')
+ CenterY = paste0('y0', paste0(' + ', colnames(dataWide)[-c(1:2)], collapse=''), ' ~ 0*1')
+ LVmodel = paste0(IModel, SModel, CenterY)
+
+ suppressWarnings({semres = growth(LVmodel, data=dataWide)})
+ varRes = sqrt(coef(semres)[ntime+1])
+ varRE = coef(semres)[c('I~~I','S~~S')]
+ covRE = coef(semres)[c('I~~S')]
+ corRE = covRE/prod(sqrt(varRE))
+ fixed = coef(semres)[c('I~1', 'S~1')]
+ list(varRes=varRes, varRE=varRE, corRE=corRE, fixed=fixed)
+}
+
+
+summarize_growth_mixed <- function(results, growth=T, whichpar='fixed', clean=F, settings=grid) {
+
+ if (!clean) fail_idx = replicate(45, rep(FALSE, 500), simplify = F)
+ if (clean) {
+ if (growth){
+ fail_idx = sapply(results, function(x)
+ sapply(x, function(res) any(is.nan(unlist(res)))), simplify=F) # | abs(res$corRE)>1
+ }
+ else {
+ fail_idx = sapply(results, function(x)
+ sapply(x, function(res) any(res$corRE==1 | res$corRE==-1)), simplify=F)
+ }
+ }
+
+ if(whichpar=='fixed'){
+ # fixed effects
+ resultFE0 = sapply(1:length(results), function(x)
+ sapply(results[[x]][!fail_idx[[x]]], function(res) res$fixed), simplify=F)
+ resultFE = lapply(resultFE0, function(res) c(rowMeans(res), c(apply(res, 1, quantile, p=c(.025,.975)))))
+ resultFE = do.call('rbind', resultFE); colnames(resultFE) = c('Int','Time', 'LL_Int', 'UL_Int', 'LL_Time', 'UL_Time')
+ resultFE = data.frame(settings, resultFE)
+ return(resultFE)
+ }
+
+ if(whichpar=='varRes'){
+ stop('Not wanting to parse residual variance for 5 to 25 scores at present. :P')
+ # residual variance
+ # resultvarRes0 = sapply(1:length(results), function(x)
+ # sapply(results[[x]][!fail_idx[[x]]], function(res) res$varRes), simplify=F)
+ # resultvarRes = lapply(resultvarRes0, function(res) c(rowMeans(res), c(apply(res, 1, quantile, p=c(.025,.975)))))
+ # resultvarRes = do.call('rbind', resultvarRes); colnames(resultFE) = c('Int','Time', 'LL_Int', 'UL_Int', 'LL_Time', 'UL_Time')
+ # resultvarRes = data.frame(settings, round(resultvarRes, 2))
+ # return(resultvarRes)
+ }
+
+ if (whichpar=='varRE'){
+ # random effects variance
+ resultvarRE0 = sapply(1:length(results), function(x)
+ sapply(results[[x]][!fail_idx[[x]]], function(res) res$varRE), simplify=F)
+ resultvarRE = lapply(resultvarRE0, function(res) c(rowMeans(res), c(apply(res, 1, quantile, p=c(.025,.975)))))
+ resultvarRE = do.call('rbind', resultvarRE); colnames(resultvarRE) = c('Int','Time', 'LL_Int', 'UL_Int', 'LL_Time', 'UL_Time')
+ resultvarRE = data.frame(settings, resultvarRE)
+ return(resultvarRE)
+ }
+
+ if (whichpar=='corRE'){
+ # cor random effects
+ resultcorRE0 = sapply(1:length(results), function(x)
+ sapply(results[[x]][!fail_idx[[x]]], function(res) res$corRE), simplify=F)
+ resultcorRE = lapply(resultcorRE0, function(res) c(mean(res, na.rm=T), quantile(res, p=c(.025,.975), na.rm=T)))
+ resultcorRE = do.call('rbind', resultcorRE); colnames(resultcorRE) = c('corRE_est', 'LL_corRE', 'UL_corRE')
+ resultcorRE = data.frame(settings, resultcorRE)
+ return(resultcorRE)
+ }
+}
+
+summary_df <- function(g, m, whichpar='fixed') {
+ require(dplyr)
+
+ if(whichpar=='fixed'){
+ # fixed effects
+ res = left_join(g, m, by=c('nClusters', 'nWithinCluster', 'corRE')) %>%
+ arrange(nClusters, nWithinCluster, corRE) %>%
+ mutate(widthGrowth_Int = UL_Int.x-LL_Int.x,
+ widthMixed_Int = UL_Int.y-LL_Int.y,
+ mixedWidthMinusgrowthWidth_Int = widthMixed_Int-widthGrowth_Int,
+ widthGrowth_Time = UL_Time.x-LL_Time.x,
+ widthMixed_Time = UL_Time.y-LL_Time.y,
+ mixedWidthMinusgrowthWidth_Time = widthMixed_Time-widthGrowth_Time)
+ return(res)
+ }
+
+ if(whichpar=='varRes'){
+ stop('Not wanting to parse residual variance for 5 to 25 scores at present. :P')
+ # residual variance
+ # growthvarRes0 = sapply(1:length(results), function(x)
+ # sapply(results[[x]][!fail_idx[[x]]], function(res) res$varRes), simplify=F)
+ # growthvarRes = lapply(growthvarRes0, function(res) c(rowMeans(res), c(apply(res, 1, quantile, p=c(.025,.975)))))
+ # growthvarRes = do.call('rbind', growthvarRes); colnames(growthFE) = c('Int','Time', 'LL_Int', 'UL_Int', 'LL_Time', 'UL_Time')
+ # growthvarRes = data.frame(grid, round(growthvarRes, 2))
+ # return(growthvarRes)
+ }
+
+ if (whichpar=='varRE'){
+ # random effects variance
+ res = left_join(g, m, by=c('nClusters', 'nWithinCluster', 'corRE')) %>%
+ arrange(nClusters, nWithinCluster, corRE) %>%
+ mutate(widthGrowth_Int = UL_Int.x-LL_Int.x,
+ widthMixed_Int = UL_Int.y-LL_Int.y,
+ mixedWidthMinusgrowthWidth_Int = widthMixed_Int-widthGrowth_Int,
+ widthGrowth_Time = UL_Time.x-LL_Time.x,
+ widthMixed_Time = UL_Time.y-LL_Time.y,
+ mixedWidthMinusgrowthWidth_Time = widthMixed_Time-widthGrowth_Time)
+ return(res)
+ }
+
+ if (whichpar=='corRE'){
+ # cor random effects
+ res = left_join(g, m, by=c('nClusters', 'nWithinCluster', 'corRE')) %>%
+ arrange(nClusters, nWithinCluster, corRE) %>%
+ mutate(widthGrowth = UL_corRE.x-LL_corRE.x,
+ widthMixed = UL_corRE.y-LL_corRE.y,
+ mixedWidthMinusgrowthWidth = widthMixed-widthGrowth)
+ return(res)
+ }
+}
+
+
+bias <- function(res, whichpar='fixed') {
+ require(dplyr)
+ if (whichpar=='fixed') {
+ out = res %>%
+ arrange(nClusters, nWithinCluster, corRE) %>%
+ mutate(biasLGC_Int = Int.x-3,
+ biasLGC_Time = Time.x-.75,
+ biasMM_Int = Int.y-3,
+ biasMM_Time = Time.y-.75)
+ return(out)
+ }
+
+ else if (whichpar=='varRE') {
+ out = res %>%
+ arrange(nClusters, nWithinCluster, corRE) %>%
+ mutate(biasLGC_Int = Int.x-1,
+ biasLGC_Time = Time.x-1,
+ biasMM_Int = Int.y-1,
+ biasMM_Time = Time.y-1)
+ }
+
+ else if (whichpar=='corRE'){
+ out = corEst %>%
+ arrange(nClusters, nWithinCluster, corRE) %>%
+ mutate(biasLGC_corRE = corRE_est.x-corRE,
+ biasMM_corRE = corRE_est.y-corRE)
+ }
+
+ return(out)
+
+}
\ No newline at end of file
diff --git a/misc_code_for_visuals.R b/misc_code_for_visuals.R
new file mode 100644
index 0000000..6755e45
--- /dev/null
+++ b/misc_code_for_visuals.R
@@ -0,0 +1,74 @@
+library(tidyverse); library(modelr); library(mgcv); library(plotly); library(lazerhawk)
+
+mod = gam(accel ~ s(times, bs='gp'), data=MASS::mcycle)
+
+ax = list(
+ title = "",
+ zeroline = FALSE,
+ showline = FALSE,
+ showticks = FALSE,
+ showticklabels = FALSE,
+ showgrid = FALSE
+)
+
+MASS::mcycle %>%
+ add_predictions(mod) %>%
+ plot_ly(x=~times, y=~accel) %>%
+ add_markers(color=~accel, colors=viridis::plasma(133), showlegend=F) %>%
+ add_lines(y=~pred, color=I('#ff5500'), opacity=.5, showlegend=F) %>%
+ hide_colorbar() %>%
+ theme_plotly() %>%
+ layout(xaxis = ax, yaxis = ax)
+
+
+input = data_frame(
+ n = 10,
+ dataMean = 7,
+ dataVar = 3,
+ priorMean = 2,
+ priorVar = 1
+)
+
+simN = 500
+theta = seq(0, 10, length.out = simN)
+
+obs = rnorm(input$n, input$dataMean, sqrt(input$dataVar))
+prior = data.frame(Distribution='Prior', theta=theta,
+ density = dnorm(theta, input$priorMean, sqrt(input$priorVar))) %>%
+ mutate(density=density/sum(density))
+
+like = data.frame(Distribution='Likelihood', theta=theta,
+ density = sapply(theta, function(parm) exp(sum(dnorm(obs, mean=parm, sd=sqrt(input$dataVar), log = T))))) %>%
+ mutate(density=density/sum(density))
+
+denom = sum(like$density*prior$density)
+post = data.frame(Distribution='Posterior', theta=theta,
+ density = like$density*prior$density/denom) %>%
+ mutate(density=density/sum(density))
+
+thetamean = sum(post$density*theta)
+plotdata = rbind(prior, like, post)
+
+
+
+
+
+g = ggplot(aes(x=theta, y=density, group=Distribution, color=Distribution, fill=Distribution), data=plotdata) +
+ geom_ribbon(aes(ymin=0, ymax=density), alpha=.5 ) +
+ geom_point(aes(x=value, y=0), data=data.frame(Distribution=c('Prior', 'Likelihood', 'Posterior'),
+ value=c(input$priorMean, mean(obs), thetamean)),
+ color=alpha('#ff5503', .25)) +
+ xlab('') +
+ lims(x=c(0, 10)) +
+ lazerhawk::theme_trueMinimal() +
+ theme(axis.title.x=element_text(color=alpha('black',.6), vjust=.1, hjust=.5),
+ axis.text.y=element_blank(),
+ axis.title.y=element_blank(),
+ axis.ticks.y=element_blank(),
+ strip.text=element_text(color=alpha('black',.5), vjust=.01),
+ legend.position='none',
+ plot.background=element_rect(fill = "transparent",colour = NA))
+
+ggplotly(g, tooltip='none') %>%
+ theme_plotly() %>%
+ layout(xaxis=ax)
diff --git a/slides.css b/slides.css
new file mode 100644
index 0000000..2728e8c
--- /dev/null
+++ b/slides.css
@@ -0,0 +1,162 @@
+@import url('https://fonts.googleapis.com/css?family=Roboto');
+
+.emph {
+ color: #ff5503;
+ font-weight: 500;
+}
+
+/* pack func and objclass colors come from hcl(seq(90,360, length.out=4), c=80, l=80)*/
+.pack {
+ color: #AC9CFF; /*#e41a1c*/
+ font-weight: 500;
+}
+
+.func { /*#984ea3;can just use `` instead*/
+ color: #00CBB6;
+ font-weight: 500;
+}
+
+.objclass {
+ color: #AAB400; /*#4daf4a; #FFC5D0*/
+ font-weight: 500;
+}
+
+
+
+
+a {
+ color: #1e90ff; /*dodgerblue*/
+}
+
+.resize{
+ width: 400px;
+ height: auto;
+}
+
+
+.reveal section p {
+ font-family: 'Roboto', sans-serif;
+ text-align: left;
+}
+
+.reveal ol,
+.reveal dl,
+.reveal ul {
+ display: block;
+ text-align: left;
+ font-size: 75%;
+ margin: 0 0 0 1em;
+
+}
+
+.reveal section img {
+ background: transparent; /*theme background color: https://github.com/rstudio/revealjs/tree/master/inst/reveal.js-3.2.0/css/theme*/
+ border: 0;
+ box-shadow: 0 0 0 rgba(0, 0, 0, 0.15);
+ /*max-width: 500px;*/
+}
+
+
+.reveal .footer {
+ position: absolute;
+ bottom: 1em;
+ left: 1em;
+ font-size: 0.5em;
+}
+
+.reveal .slides section .fragment {
+ font-size: 75%;
+ font-weight: 600;
+}
+
+
+.reveal .progress span {
+ display: none;
+}
+
+.reveal .playback {
+ display: none;
+}
+
+.reveal .controls {
+ display: none !important;
+}
+
+.reveal pre code {
+ box-sizing: border-box;
+ left: 0; /* This changes where the code chunk box actually starts */
+ /* padding: 10px 0 10px 60px; /* Change the last value here to move the text left or right */
+ position: relative;
+ width: 100%; /* This changes where the code chunk box ends on the right side */
+ font-size: 100%; /*changes output size; and comments*/
+ box-shadow: .5 rgba(0, 0, 0, 0.15);
+}
+
+.col2 {
+ columns: 2 200px; /* number of columns and width in pixels*/
+ /*font-size: 50%;*/ /* ignored, but can do in the div style within a slide*/
+ /*font-size: 20pt;*/
+ -webkit-columns: 2 200px; /* chrome, safari */
+ -moz-columns: 2 200px; /* firefox */
+}
+
+.col2 .reveal pre {
+ font-size:25%;
+}
+
+.col3 {
+ columns: 3 100px;
+ /*font-size: 16pt;*/
+ -webkit-columns: 3 100px;
+ -moz-columns: 3 100px;
+}
+
+
+/* none of the following seemed to do much*/
+.title {
+ text-align: center;
+ font-variant: small-caps;
+ margin: 0 auto;
+
+/* background-image: url('img/CSCAR.png');
+ background-position: center center;
+ background-attachment: fixed;
+ background-repeat: no-repeat;
+ background-size: 100% 100%; */
+}
+
+
+.sectionSlide {
+ color: #1e90ff;
+ text-align: center;
+ font-size: 200%;
+ font-variant: small-caps;
+ font-weight: 500;
+ background: linear-gradient(to bottom, #002b36, white);
+}
+
+.reveal .sectionSlideTitle {
+ data-background-color: white;
+}
+
+#linearBg2 { /* fallback */
+ background-color: #002b36;
+ background-repeat: repeat-x; /* Safari 4-5, Chrome 1-9 */
+ background: -webkit-gradient(linear, 0% 0%, 0% 100%, from(#002b36), to(#2F2727)); /* Safari 5.1, Chrome 10+ */
+ background: -webkit-linear-gradient(top, #2F2727, #002b36); /* Firefox 3.6+ */
+ background: -moz-linear-gradient(top, #2F2727, #002b36); /* IE 10 */
+ background: -ms-linear-gradient(top, #2F2727, #002b36); /* Opera 11.10+ */
+ background: -o-linear-gradient(top, #2F2727, #002b36);
+}
+
+html.opac75 .slide-background.present {
+ opacity: 0.75;
+}
+
+html.opac50 .slide-background.present {
+ opacity: 0.50;
+}
+
+html.opac25 .slide-background.present {
+ opacity: 0.25;
+}
\ No newline at end of file
+
+
+
+
+
+
+
+
+
+
+
+Statistical
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ cscar.research.umich.edu
+Data Science Collaboration
+
+
+
Statistical
and
Computational Programming
+
+Data Science Serenity
+
+