You are analyzing a data set that includes the number of visitors to a website per day. How would you categorize this data type?
- Continuous data
- Discrete data
- Nominal data
- Ordinal data
The number of visitors to a website per day would be discrete data as it is countable in a finite amount of time.
For data with outliers, the _____ is typically a better measure of central tendency as it is less sensitive to extreme values.
- Mean
- Median
- Mode
- Variance
The "Median" is less sensitive to extreme values, or outliers, in a dataset. Therefore, it's often a better measure of central tendency when outliers are present.
If you are working with a large data set and need to produce interactive visualizations for a web application, which Python library would be the most suitable?
- Bokeh
- Matplotlib
- Plotly
- Seaborn
Plotly is well-suited for creating interactive visualizations and can handle large data sets efficiently. It also supports rendering in web applications, making it ideal for this scenario.
What type of bias could be introduced by mean/median/mode imputation, particularly if the data is not missing at random?
- Confirmation bias
- Overfitting bias
- Selection bias
- Underfitting bias
Mean/Median/Mode Imputation, particularly when data is not missing at random, could introduce a type of bias known as 'Selection Bias'. This is because it might lead to incorrect estimation of variability and distorted representation of true relationships between variables, as the substituted values may not accurately reflect the reasons behind the missingness.
How does the number of imputations affect the accuracy of multiple imputation?
- More imputations, less accuracy
- More imputations, more accuracy
- Number of imputations doesn't affect accuracy
- Only one imputation is needed for full accuracy
The number of imputations directly affects the accuracy of multiple imputation. More imputations result in more accurate estimates, up to a point. Although the exact number depends on the proportion and nature of the missing data, often 20 to 100 imputations are recommended in the literature.
In data analysis, EDA stands for _______.
- Empirical Data Assessment
- Exploratory Data Analysis
- Exponential Data Analysis
- Expressive Data Assimilation
In data analysis, EDA stands for Exploratory Data Analysis. It is an approach to analyzing data sets to summarize their main characteristics, often with visual methods.
Can multiple imputation be applied when data are missing completely at random (MCAR)?
- No
- Only if data is numerical
- Only in rare cases
- Yes
Yes, multiple imputation can be applied when data are missing completely at random (MCAR). In fact, it is a flexible method that can be applied in various missing data situations including MCAR, MAR (missing at random), and even NMAR (not missing at random).
You're in the 'explore' phase of the EDA process and you notice a potential error back in the 'wrangle' phase. How should you proceed?
- Conclude the analysis with the current data.
- Go back to the wrangling phase to correct the error.
- Ignore the error and continue with the exploration.
- Inform the stakeholders about the error.
If you notice a potential error in the 'wrangle' phase while you are in the 'explore' phase, you should go back to the 'wrangle' phase to correct the error. Ensuring the accuracy and quality of the data during the 'wrangle' phase is crucial for the validity of the insights drawn in subsequent phases.
What is the impact on training time if missing data is incorrectly handled in a large dataset?
- Decreases dramatically.
- Depends on the specific dataset.
- Increases dramatically.
- Remains largely the same.
If missing data is not handled correctly, particularly in a large dataset, the training time can increase significantly. This is because the model might struggle to learn from the distorted data, requiring more time to try to fit the data.
The _______ method of feature selection involves removing features one by one until the removal of further features decreases model accuracy.
- Backward elimination
- Forward selection
- Recursive feature elimination
- Stepwise selection
The backward elimination method of feature selection involves removing features one by one until the removal of further features decreases model accuracy. This process starts with a model trained on all features and iteratively removes the least important feature until the overall model performance declines.