CISSE Advanced Level
Big Data Analytics flashcards
Read the question, say the answer to yourself, then turn the card to check. The first 10 of 62 cards are free; the rest come with the subject.
Card 1 of 62
Jump to a card
Complete Flashcard Reference
Review every card in this deck. Open any card to reveal its answer. The first 10 are free; the other 52 come with the subject.
What are the five V's of big data?
Volume: the sheer scale of the data. Velocity: the speed at which it is generated and must be processed. Variety: the many forms it arrives in, structured, semi-structured and unstructured. Veracity: its trustworthiness and quality. Value: the insight or decision analysis turns it into.
What is the gradient descent update rule, and what does the learning rate control?
new w = old w - (learning rate x gradient). The parameter moves against the gradient, the direction that reduces the loss fastest, and the learning rate sets the size of each step. Repeated many times, it is how most models in the paper are fitted.
How does Dask differ from Pandas, and why does its API mirror Pandas'?
Pandas manipulates structured data in a DataFrame within one machine's memory. Dask extends the same DataFrame-style interface to data too large for one machine, splitting it into partitions processed in parallel across cores or machines. The API mirrors Pandas so existing Pandas skills and code carry over to the large-scale case.
What are the two ways of handling missing values?
Deletion: drop the affected row or column, defensible where little is missing. Imputation: fill the gap with the mean, median, mode, or a value predicted from other features. The median is the robust choice where extreme values remain.
How does betweenness centrality differ from degree centrality?
Degree centrality: the number of edges connected to a node, how directly connected it is. Betweenness centrality: how often a node lies on the shortest path between two other nodes. A node with a low degree can still have high betweenness if it is the bridge between otherwise separate parts of the network.
What are the two steps of a K-means iteration?
1. Assign each point to its nearest current centroid. 2. Recompute each centroid as the mean of the points now assigned to it. The steps repeat until the clusters stop changing. k must be chosen in advance, commonly by the elbow method.
How are support, confidence and lift calculated for a rule A -> B?
support(A) = count(A) / N, the share of all transactions containing A. confidence(A->B) = support(A,B) / support(A). lift(A->B) = confidence(A->B) / support(B). A lift above 1 shows a genuine positive association.
Which five figures does a box plot summarise?
Minimum, first quartile, median, third quartile and maximum. Points beyond a defined range are plotted individually as outliers, and several groups can be compared side by side compactly.
Which principles of data processing does section 25 of Kenya's Data Protection Act, 2019 set out, as grouped in the guide?
Lawfulness, fairness and transparency. Purpose limitation and data minimisation. Accuracy and storage limitation. Accountability, including being able to explain the basis of a consequential decision.
How does collaborative filtering differ from content-based filtering, and what is the cold-start problem?
Collaborative filtering recommends from the preferences of similar users, without understanding the items. Content-based filtering recommends items similar in content or attributes to what a user has liked. Cold start: a brand-new item has no interaction history, so collaborative filtering cannot recommend it. Hybrids combine both to cover each other's weakness.
How are artificial intelligence, machine learning and deep learning related?
This answer comes with Big Data Analytics. Buy the subject
Which opportunity answers each of big data's four core challenges?
This answer comes with Big Data Analytics. Buy the subject
What are scalars, vectors, matrices and tensors in terms of rank?
This answer comes with Big Data Analytics. Buy the subject
When can two matrices be multiplied, and how is each entry of the product found?
This answer comes with Big Data Analytics. Buy the subject
What is an eigenvector, and why does it matter in big data analytics?
This answer comes with Big Data Analytics. Buy the subject
What does a zero determinant tell you, and what does SVD do?
This answer comes with Big Data Analytics. Buy the subject
Why does least squares square the errors before summing them?
This answer comes with Big Data Analytics. Buy the subject
What is the difference between a predictor and a target?
This answer comes with Big Data Analytics. Buy the subject
Why should an outlier not automatically be deleted?
This answer comes with Big Data Analytics. Buy the subject
How does feature transformation differ from feature creation?
This answer comes with Big Data Analytics. Buy the subject
What makes a graph directed, undirected or weighted?
This answer comes with Big Data Analytics. Buy the subject
What do closeness centrality and the clustering coefficient measure?
This answer comes with Big Data Analytics. Buy the subject
Why is a graph database faster than a relational one for deep chains of relationships?
This answer comes with Big Data Analytics. Buy the subject
What is Neo4j, and what is Cypher for?
This answer comes with Big Data Analytics. Buy the subject
What are the stages of the machine learning pipeline?
This answer comes with Big Data Analytics. Buy the subject
How do feature extraction, feature selection and labelling differ?
This answer comes with Big Data Analytics. Buy the subject
Why must a model be evaluated on data it did not train on?
This answer comes with Big Data Analytics. Buy the subject
How does k-fold cross-validation differ from a single train-test split?
This answer comes with Big Data Analytics. Buy the subject
What does each library in the core Python data science stack do?
This answer comes with Big Data Analytics. Buy the subject
Why is dimension reduction often applied before clustering?
This answer comes with Big Data Analytics. Buy the subject
How does agglomerative clustering differ from divisive clustering?
This answer comes with Big Data Analytics. Buy the subject
When is K-means preferred to hierarchical clustering, and the other way round?
This answer comes with Big Data Analytics. Buy the subject
How does logistic regression differ from linear regression?
This answer comes with Big Data Analytics. Buy the subject
What is entropy in a decision tree, and how is it calculated for a binary node?
This answer comes with Big Data Analytics. Buy the subject
How does a random forest reduce the overfitting of a single decision tree?
This answer comes with Big Data Analytics. Buy the subject
In one line each: what do SVM, Naive Bayes and KNN classification do?
This answer comes with Big Data Analytics. Buy the subject
How does Linear Discriminant Analysis differ from PCA as a dimension reduction technique?
This answer comes with Big Data Analytics. Buy the subject
Which model suits a bank that must explain each credit decision to the applicant, and why?
This answer comes with Big Data Analytics. Buy the subject
Which property lets the Apriori algorithm avoid testing every item combination?
This answer comes with Big Data Analytics. Buy the subject
Why can a rule with high confidence still be a poor association?
This answer comes with Big Data Analytics. Buy the subject
What does a single artificial neuron compute?
This answer comes with Big Data Analytics. Buy the subject
What are sigmoid, ReLU and softmax used for in a neural network?
This answer comes with Big Data Analytics. Buy the subject
When would you use a CNN, and when an RNN?
This answer comes with Big Data Analytics. Buy the subject
What are the text analytics steps that prepare text for a model?
This answer comes with Big Data Analytics. Buy the subject
How does sentiment analysis differ from topic modelling?
This answer comes with Big Data Analytics. Buy the subject
Which chart suits a trend, a comparison of categories, and a share of a whole?
This answer comes with Big Data Analytics. Buy the subject
What do a correlation heatmap, a biplot and a word cloud each show?
This answer comes with Big Data Analytics. Buy the subject
How can a model be biased without using a protected characteristic as a feature?
This answer comes with Big Data Analytics. Buy the subject
Why is compliance with the Data Protection Act not enough on its own?
This answer comes with Big Data Analytics. Buy the subject
What happened in Regus Kenya Limited v Data Protection Commissioner (2025)?
This answer comes with Big Data Analytics. Buy the subject
On what three things does Python's dominance in big data analytics rest?
This answer comes with Big Data Analytics. Buy the subject
A telecom firm receives millions of call detail records an hour, in several formats from different equipment vendors, some with missing fields. Which V's does this illustrate?
This answer comes with Big Data Analytics. Buy the subject
How does the elbow method help choose k for K-means?
This answer comes with Big Data Analytics. Buy the subject
How should a discovered association rule be validated before it drives a business action?
This answer comes with Big Data Analytics. Buy the subject
What are four applications of association rule mining?
This answer comes with Big Data Analytics. Buy the subject
What does 'deep' refer to in deep learning, and why did it become practical only recently?
This answer comes with Big Data Analytics. Buy the subject
In a feedforward network, what is the trade-off in choosing the number of hidden layers and neurons?
This answer comes with Big Data Analytics. Buy the subject
Why does social media analytics draw on social network analysis as well as text techniques?
This answer comes with Big Data Analytics. Buy the subject
What weakness of content-based filtering does a hybrid recommender system cover?
This answer comes with Big Data Analytics. Buy the subject
How does a kernel density estimate differ from a histogram?
This answer comes with Big Data Analytics. Buy the subject
How is the result of a clustering algorithm typically visualised, and why?
This answer comes with Big Data Analytics. Buy the subject
Besides the Data Protection Act, 2019, which other Kenyan rules bear on a big data analytics project?
This answer comes with Big Data Analytics. Buy the subject
How to revise with these cards
- Answer out loud or on paper before you turn the card: recalling is what fixes it.
- Go through the deck in short sessions of ten to fifteen minutes, every day.
- Shuffle once you know the order, so you are not answering from memory of the sequence.
- When a card catches you out, read the chapter it links to, then try the topic in a past paper.