Data Science - MODULE 1

Question 1

Data science is where three domains meet?

Answer 1

1. Applied mathematics and statistics
2. Domain knowledge
3. Computer science

Question 2

Die 4 hoof libraries wat ons gaan gebruik

Answer 2

Numpy
Pandas
Scikit-learn
Matplotlib

Question 3

When do we use statistical learning?

Answer 3

When the pattern cannot be observed directly

Question 4

Inputs are commonly referred to as?

Answer 4

Predictors or features

Question 5

Outputs are also called?

Answer 5

Responses

Question 6

Three spheres of statistical learning? Which ones will we deal with?

Answer 6

Supervised, unsupervised, semi-supervised.
We will work with the first two

Question 7

Die doel van unsupervised learning?

Answer 7

NIE om n spesifieke reaponse re predict nie, maar eerder om patrone te soek

Question 8

Daai hele video oor hoefdings inequality, gaan oor hoeveel observasies jy nodig het om onder n seker error te kom

Answer 8

P|abs(v-u)>e| <=2exp(-2(e^2)N)

Question 9

Wat is die eerste vraag om te bepaal wat se model gebruik moet word?

Answer 9

Is daar n spesifieke response variable. I dien JA, supervised. Nou kan die Neurale netwerke of tree-based model wees

Question 10

Groot verskil tussen neurale netwerke en tree-based models

Answer 10

Neurale netwerke is gebou vir akkuraatheid, en dit is n swart boks. So waar jy moet kan bewys hoekom n seker besluit geneem is, is dit nie rerig n opsie nie. Tree-based models is interpretablr, maw ons kan sien hoe daar by n seker model uitgekom is

Question 11

Hoof verskil tussen k-means en hierarchical clustering?

Answer 11

K-means kies jy die aantal groepe. Hierarchical, word dit self ontwikkel. So waar jy baie data het, is dit soms beter om k-means te gebruik

Question 12

Hoekom kan k-means clustering partykeer nie akkuraat wees nie?

Answer 12

Prefers spehrical clusters?

Question 13

Wanneer daar responses is, en dit is diskrete data punte - watse tipe leer sal dit behels

Answer 13

Tree-based classification methods