WEEK 10

Question 1

Seasonality is

Answer 1

defined by the tendency of time series data to exhibit behaviour that repeats itself every k period

to identify the value is to plot your time series data and observe patters

Question 2

examples of seasonality aree

Answer 2

daily - city traffic
weekly-rental car demand
monthly/quarterly - online spending
yearly - temperature

Question 2

additive seasonality

Answer 2

happens when the variation in data is due to seasonality is constant over time, this can draw parallel lines at peaks and bottoms

Question 3

multiplicative seasonality occurs

Answer 3

when the variation due to seasonality either increases or decreases over time, in this case you will have non parallel lines at the peaks and trough’s

Question 4

the model for additive is called additive decomposition model

Answer 4

y(under t) = trend (under t) + seasonality (under t) + cyclic (under t) + Randomness (under t) = T(under t) +S(under t) +C+ e (under t)

only look at short time then we assume that cyclical fluctuations do not exist so we don’t add it

Question 5

last step of forecast is to

Answer 5

namely make prdiction based on the past time serties and use the model to forecast we can only estamte the T AND S they are also time depended and not fixed constants cuz they both change over time so we dont obtain one specific value (using CMAs OR SMAs we can find trend)

Question 6

seasonality values s CALCULATION DE TRENDED time series excel Q2-($Q$9/7)

Answer 6

D(under t) = (initial) y(under t) (- assumed)T (under t), so the actual value minus the trend

We group the values, like if there are 3 weeks we do Mondays together, Tuesdays together, etc, and calculate the average of all the days AvgD(under i) = sum of all (under i) D(under i) / n (under i)

then s (under i) = AvgD(under i) - sum of all (under i) D(under i) / k

Then take averages per day minus the average of all the averages divided by 7

Question 7

for the multiplicative model we use multiplicative decompositions model

Answer 7

y(t under) = T (t under) x S (t under) x C (t under) x e (t under)

WE DO AGAIN without cyclic

Question 8

what are forecast errors

Answer 8

error = actual value - predicted value

Question 9

we want ti know for the forecast error

Answer 9

how bis the misatke is and how ofter we make em

Question 10

we calculate MAE (MEA ABSOLUTE ERROR) avarage size of the misatkes

Answer 10

MSE (mean squared error) - average of mistakes squared (punishes big mistakes more)

Question 11

RMSE (root mean squared error) - like MSE but goes back to the same units as the data

Answer 11

MAPE (mean absolute percentage error) mistakes as % of the real value

Question 12

if there is a big distance in there error it means that the randomness is

Answer 12

is very high

Question 13

Steps to calculate the errors

Answer 13

Use the formula, then take the absolute value of each error (ignore the minus signs)

square each error (multiply by itself)

find averages
-MSE - AVERAGE OF SQUARED ERRORS
-MAE - AVARAGE OF ABSOLUTE ERROS

Take the square root of MSE and find RMSE

Find MAPPE by dividing the absolute error by the real value and turning it into a percentage %

Question 14

In Excel, do trend first, then seasonality. We paste for each day of the week

Answer 14

and to predict we do t +s ad then extend that

to find the value, we do period litres and predict, and insert a chart to plot the forecast, watch week 10 vle if I don’t get

Question 15

Data analytics

Answer 15

Means using tech uques algorithm and technologies to discover patterns trends and

Question 16

Traditional analysis

Question 17

Big data analytics

Answer 17

Massive messy and fast data machine learning AI 4eal time dashboard, neutral networks decision trees clustering, instant or dynamic analysis
Tools hadoop, spark, python

Question 18

Big data

Answer 18

Refers to data sets tjat are so large or complex tjat traditional data processing applications are inadequate

Question 19

5 Vs fi Big data

Answer 19

Volume : massive amounts of data

Velocity : data generated quickly

Variety : different types and fonts (text imagine or video)

Veracity : uncertainty and quality issue in data

Value: potential to generate insight and economic value

Question 20

Future implications

Answer 20

Innovation and competition : as technology evolves big data will play even more critical role in fostering innovation and competitive advantage across all sectors

Career opportunities : there is a growing demand for professionals skilled in data analytics and data driven decision making across various industries

Ethical and social implications must consider the ethical implications of data use including privacy concerns and data clsedurry as these will be significant aspects of future regulatory and business environments

Question 21

Tools for big data

Answer 21

Marketing analytics such as Google analytics or adobe analytics

Machine learning analytics

Data visualisation tool such as Hadoop or apache Cassandra

Question 22

Origins

Answer 22

From early databases to the explosion of data in the digital age, highlight key milestones like the development of the internet social media

Question 23

Big data has become essential for competitive advantage

Answer 23

Enabling business to make informed decision predict trends and personalise customer experience

Question 24

Skills needed for careers in big data

Answer 24

Data literacy - ability to understand and interpret data Statistical knowledge - understand concepts such as correlation and regression Programming - Python and R Critical thinking - turning data into strategic decisions Communication skills - explaining data insights clearly to ithers

Question 25

Data analyst

Answer 25

Clean process and find insights from datasets

Question 26

Business intelligence analyst

Answer 26

Use data to support business decision

Question 27

Data scientist

Answer 27

Build advanced models to predict future trends