Halpin Analytics BUAL2600 FINAL

5.0 (1 review)
Steps of the analytics process
Click the card to flip 👆
1 / 45
Terms in this set (45)
Steps of the analytics process
business problem, choose data sourcing method, collect descriptives, make predictions, ask next question
continuous vs. discrete variables
continuous- fraction/decimal
discrete- integer

*given a variable, be able to identify if it is continuous or discrete
sampling methods: convenient
readily available data, chosen because it's easy/convenient to get the data
example: using all customer's from today's sales
*biased
sampling methods: voluntary
self-selected sample, those who want to participate do so
ex: survey
*voluntary sampling is ALWAYS biased
sampling methods: simple (simple random sampling)
subset of individuals chosen from a larger set, each individual is chosen randomly and entirely by chance
sampling methods: stratified (stratified random sample)
population is divided into non-overlapping groups (called strata), then select a simple random sample from each strata
ex: dividing males and females, states/regions, etc.
sampling methods: systematic
list population, select random starting point, sample each nth element
sampling methods: cluster
divide population into clusters, randomly select clusters and sample everyone/everything in that cluster
sampling biases: under coverage/coverage bias
not all of the population was ever on the "list of consideration" to be sampled, improper representation of the whole population
sampling biases: sampling error
choosing the wrong sampling method
sampling biases: measurement errorsrecording the wrong responsewhen to use graphs: bar graphused with qualitative data, bars DO NOT touchwhen to use graphs: histogramused with quantitative data, bars do touchwhen to use graphs: pie chartused to show percentage/part of a wholewhen to use graphs: scatterplotused with TWO quantitative variables and its purpose is to show the relationship between the variablesdata sourcing methods: point of saleinformation collected at the store, receiptsdata sourcing methods: clickstreamtracking online navigation, Amazon/Googledata sourcing methods: social mediainformation shared on social mediadata sourcing methods: sensorsdevice that tracks environmentcontent analysis (when to use it and purpose)researching and analyzing QUALITATIVE datar (correlation coefficient)/identifying relationships between quantitative variablesstrength is based on how close points are to linear form on a scatterplot stronger relationship is closer to one, weaker relationship is closest to 0 ex: -.021 is weak, .951 is stronghow to find a relative frequencyfrequency is proportion or part of a whole example: 13 out of 40 participants are males 13/40 = .325, so relative frequency of male participants is .325data visualization vs. visual analyticsdata visualization- data being displayed visual analytics- adding prediction to thisparts of a graph *be able to identify parts and what would make the graph misleadingtitle, axis labels, scale, key, data valuespurpose of a graphto tell an accurate storyparts of a consent formpurpose, confidential/anonymous, expectations of participant, voluntary/involuntary, risk, contact for questionsethics principles *given a scenario, be able to identify scenario usedownership, consent, privacy, openness, currencyreport vs dashboard *given a scenario, know when to use eachreport- static, historic data, snapshot, like a Powerpoint dashboard- dynamic, interactive, up-to-date, real-time updates, special software like Tableaugiven a graph, box plot, scenario or mean/median values, be able to recognize if approximately normal, skewed left, or skewed rightskewed right when: mean is larger than median, median is larger than mode skewed left when: mean is smaller than median, median is smaller than mode approximately normal when graph is symmetrical link to pictures: https://www.google.com/url?sa=i&url=https%3A%2F%2Fwww.statology.org%2Fleft-skewed-vs-right-skewed%2F&psig=AOvVaw2s7RM-jhpUSyTt0yxfrueg&ust=1619552836893000&source=images&cd=vfe&ved=0CAIQjRxqFwoTCIDrstHWnPACFQAAAAAdAAAAABADbe able to find mode, Q1, Q2, Q3, and IQR of datasetQ1- 25th percentile Q2- 50th percentile/median Q3- 75th percentile IQR= Q3-Q1 mode- appears most frequently in data setdescriptives and excel commands to find themmean =average median =MED mode =mode standard deviation of population =STDEV.P standard deviation of sample =STDEV variance =VAR.Puse population mean and standard deviation to find a z-scoreZ= (x-mean)/standard deviation (x is your value) z-score shows us the number of standard deviations that a value is from the meanfind dataset value that lies 1,2, or 3 standard deviations above or below the mean1 SD above: mean + standard deviation 1 SD below: mean- standard deviation 2 SD above: mean + (standard deviation x 2) 2 SD below: mean - (standard deviation x 2) 3 SD above: mean + (standard deviation x 3) 3 SD below: mean - (standard deviation x 3)know the 68, 95, 99 rule for finding the percent of values between -3 and 3 standard deviations68% of data lies within 1 standard deviations above/below the mean 95% is within 2 standard deviations 99% is within 3 standard deviationsknow when to use mean or median and range or IQRmedian and IQR are not affected by outliers so use this when data is skewedknow when to choose to use percent vs. percentilepercent is part of a whole, percentile is a positioning of where something is in the data set percentile: the percentage of the score in the distribution that are equal to or less than the ranking ex: 80th percentile, 80% of values lay below this point, 20% of values lay above this pointgiven dataset, find percentile value (find datapoint at a given percentile)1. put dataset in increasing numerical order 2. calculate the index(i). index is the position of the point. Index=(p/100) x n. p= percentile, n=sample size if Index is not an integer, you need to round UP if index is an integer, find average of value at index and index+1 LECTURE ON MARCH 16--20 MINUTES INgiven box plots, be able to describe dataset or compare box plotsparts of box plot Q1, Q2, Q3, min, max (anything beyond min and max is outlier) compare skewness size of our box is IQR. small box=small standard deviation, large box=larger standard deviationbe able to find outliers OR find that there are NO outliersany number greater than Q3 + (1.5 x IQR) is an outlier any number less than Q1 - (1.5 x IQR) is an outlierfind sample proportionsample proportion= count of successes/size of population count of successes refers to what you're looking for. ex: of 200 people that entered a store, 78 of them purchased something. 78/200= .39 sample proportion is .39Find probabilities less than, greater than, or between zscores using normal distributionuse z-score with normal distribution table. number listed in normal distribution table shows the probability that a z is LESS THAN the z you are using. do 1-the number in the table to find probability z is GREATER THAN the z you are using. for between, you subtract the probability of lower z-score from probability of greater z-score LECTURE FROM APRIL 6Find probabilities if given the raw score (x) using normal distributionmust convert x to z-score to find probability. z= (x-mean)/standard deviationFind 90%, 95%, and 99% confidence intervals for proportions. *formula will be given, but know how to use ituse the formula to find upper and lower limits parts of the equation you should understand so you can plug it in: p-hat : sample proportion z a/2 : standardized factor based on confidence interval (given) n : sample size APRIL 15 at 10 minutes (can't write formula on quizlet)Find the margin of error *know how to find using confidence interval formulamargin error is the right of the equation-- everything to the right of the +or-Identify the interpretation of a confidence intervalconfidence interval means probability that the estimated interval will actually contain the population parameter