Accounting Data Analytics- Questions
and Answers
A Target is: an expected attribute that we want to evaluate. (Fraud score, Interest rate)
A class is: A manually assigned category applied to a record based on an event.
(Accept/Reject)
Unsupervised approach: Don't have a specific question
- Clustering
- Co-Occurance grouping
- Profiling
- Data Reduction
Supervised approach: Trying to predict a future outcome based on historical data
- Classification
- Similarity Matching
- Regression
- Link prediction
- Causal modeling
, Decision Trees Used to divide data into smaller groups
Qualitative Data Nominal Data (color)
Ordinal Data (GPA)
Proportion (%'s)
Quantitative Data Ratio (0 matters)
Interval (0 doesn't matter)
Discrete (whole)
Continuous (decimals)
Distributions (mean)
Exploratory visualization Used to gain insights while you are interacting with data
What questions relate to scale and increments? How much data do you need to show?
What do you do with outliers?
What is the baseline?
Would contect or reference lines make the scale more meaningful?
What questions should you ask about color? What do the colors mean?
Should red be used?
What color schemes work better for color-blind?
and Answers
A Target is: an expected attribute that we want to evaluate. (Fraud score, Interest rate)
A class is: A manually assigned category applied to a record based on an event.
(Accept/Reject)
Unsupervised approach: Don't have a specific question
- Clustering
- Co-Occurance grouping
- Profiling
- Data Reduction
Supervised approach: Trying to predict a future outcome based on historical data
- Classification
- Similarity Matching
- Regression
- Link prediction
- Causal modeling
, Decision Trees Used to divide data into smaller groups
Qualitative Data Nominal Data (color)
Ordinal Data (GPA)
Proportion (%'s)
Quantitative Data Ratio (0 matters)
Interval (0 doesn't matter)
Discrete (whole)
Continuous (decimals)
Distributions (mean)
Exploratory visualization Used to gain insights while you are interacting with data
What questions relate to scale and increments? How much data do you need to show?
What do you do with outliers?
What is the baseline?
Would contect or reference lines make the scale more meaningful?
What questions should you ask about color? What do the colors mean?
Should red be used?
What color schemes work better for color-blind?
