Help
Options
Didn't know it?
click below
click below
Knew it?
click below
click below
Don't Know
Remaining cards (0)
Know
retry
shuffle
restart
0:00
Embed Code - If you would like this activity on your web page, copy the script below and paste it into your web page.
Normal Size Small Size show me how
Normal Size Small Size show me how
CSI 343 Exam 1
| Question | Answer |
|---|---|
| What is Data Science | cross-disciplinary practice that draws on methods from data engineering, descriptive statistics, data mining, machine learning, predictive analysis |
| What are the stages of a data science project? | Get, Explore, Clean, Visualize, Train & Test |
| What the approach when data is expensive | Define goal/problem first, then explore data |
| What the approach when data is cheap? | Explore data first, then see how it can be used |
| What is a Primary Data source? | Data collected directly by a resarcher |
| Example of a primary data source | HMDA, Census |
| What is a Secondary Data source? | Data collected by someone else |
| Secondary Data source example | Kaggle, UCI ML Repository, Data.gov |
| Problems to look for when exploring data | Missing/unusual values, anomalies/outliers, lack of diversity/representation, bias |
| How can missing values be handled? | imputing with measures of central tendency (mean,median,mode) |
| Why are unbalanced distributions problematic? | bias models, especially classification models, making them less effective |
| Difference between correlation and causation | correlation is a relationship/pattern between two variables and causation is the principle where one event causes another |
| What i a spurious correlation | Appears to be correlated but has no causation |
| Three main types of machine learning | Supervised, unsupervised, reifnorcement |
| What is supervised learning | learning from labeled data |
| What is unsupervised learning | learning from finding patterns in unlabeled data |
| What is reinforcement learning | learns decisions via environment interaction |
| What is Qualitative Data | Data that cannot be measured numerically; sorted by categories. |
| Examples of Qualitative Data | Audio, Images, text, race |
| What is Nominal Data | Categorical data with no order or inherent ranking |
| Nominal Data example | sex, race, hair color |
| What is ordinal data | Categorical data with ordering, but differences between ranks may not be equal |
| Ordinal Data example | rankings, letter grades, satisfaction levels |
| What is Quantitative Data | Numerical data that can be measured or counted |
| Quantitative data example | heigh, weight, income |
| What is Continuous Data? | Quantitative data that can take any value (measured) |
| Continuous data example | temperature, GPA |
| What is discrete data? | Quantitative data that takes specific countable values |
| Discrete data example | number of dogs, birth year |
| What is structured data? | Data prganized into rows / columns in relational databases; about 20% of enterprise data; easier to manage |
| Structured Data example | numbers, dates, strings |
| What is Unstructured Data? | Data not stored in relational databases, lack schema; 80% of enterprise data |
| Unstructured Data example | NoSQL databses (MongoDB,Couchbase,Cassandra) |
| Define Mean | Average value of a dataset |
| Arithmetic mean formula | sum of values / number of values |
| Weighted mean | Accounts for different weights of values: sum(val x weight)/sum(weight) |
| median | middle value of dataset when ordered |
| mode | values that occur most frequently |
| Unimodal | 1 mode |
| Bimodal | 2 modes |
| Multimodal | 2+ modes |
| Range | max-min |
| Percentile | value below which a percentage of data falls |
| Percentile example | 95th percentile= better than 95% of test takers |
| IQR | Interquartile Range; 75th percentile-25th percentile |
| Variance | Average squared deviation from mean; measures spread |
| Difference between population vs sample variance | population uses n in denominator; sample uses n-1 |
| Why is standard deviation easier to interpret than variance? | It is in the same units as the data (sqrt of variance) |
| Experiment | Repeatable procedure th defined otucomes |
| Experiment example | Tossing a coin |
| Sample space | set of all possible outcomes |
| Event | Subset of sample space |
| Simple event | one outcome |
| Compount event | multiple outcomes |
| mutually exclusive events | events that cannot occur together (disjoint) |
| 3 axioms of probability | Non-negativity, Unity, Additivity |
| Non-Negativity | P(A) ≥ 0 |
| Unity | P(S) = 1 |
| Additivity | If A and B are disjoint, P(A ∪ B) = P(A) + P(B) |
| State the complement rule | P(not A) = 1 – P(A). |
| State the general addition rule. | P(A ∪ B) = P(A) + P(B) – P(A ∩ B). |
| What is conditional probability | P(A|B) = probability of A given B. |
| Define dependent vs independent events. | Dependent: probability of one affects the other. Independent: no effect. |
| What does exhaustive mean in probability? | Events that cover the entire sample space (one must happen). |
| Define outlier. | Extreme observation differing from the rest. Identified by >1.5×IQR or >2 SD from mean. |
| Define anomaly | Unexpected data pattern; may indicate fraud, system failure, or novel events. |
| How to handle missing values? | Impute (mean, median, mode), create new fields, or drop column if mostly missing. |
| How to handle invalid values? | Drop them, replace with valid values, or create sentinel values. |
| MNIST dataset (28x28 grayscale digits) → What type of data? | Nominal. |
| Fashion MNIST dataset (clothing images) → Data type? | Nominal. |
| Debt-to-income ratio as intervals/values → Data type? | : Ordinal (mode but not mean). |
| Example when it’s okay to remove an entire column? | When it isn’t relevant (e.g., phone number when analyzing grades). |
| Example when it’s okay to remove an entire row? | When data is mostly missing or invalid. |
| What does an income dataset missing values illustrate? | Mean of entire dataset may not be best imputation due to regional variation. |
Created by:
user-1979725