Scribd
Upload
52 views

Explaining simulating weather forecast code using Markov chain

Uploaded by

lquibee8
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
52 views

Explaining simulating weather forecast code using Markov chain

Uploaded by

lquibee8
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 2

2.

Weather Forecast [1]


Markov Chains are used to calculate for the probabilities of events express in the form of
transition matrix. For example:
Trasition matrix: Correspond to next day
Current states Sunny Rainy Cloudy
Sunny 0.55 0.15 0.30
Rainy 0.35 0.45 0.20
Cloudy 0.30 0.40 0.30

If today is a sunny day, an average of 55% that the next day is another sunny day, or 30% with
cloudy day and only 15% that the next day is a rainy day. Besides, if it’s a rainy day today, then
45% chance that the next day will be another rainy day, 20% for cloudy day and 35% chance that
it will be a sunny day. And if today is cloudy day, then 30% for the same cloudy next day, 40%
for rainy and 30% for sunny.
In our weather forecast example, we define S = [Sunny Rainy Cloudy]. Therefore, assumed that
today is a sunny day, we define S0 = [1 0 0] because there is hundred percent of a sunny day and
zero chance of a rainy day and cloudy day.
Therefore, the next state, S1, we can calculate the matrix product S1 = S0P.
[0.55¿0.15¿0.30 ]
[1 0 0 ] . [0.35¿0.45¿0.20 ] = [0.55 0.15 0.30 ]
[0.30¿0.40¿ 0.30]
Since the formula for computing successive state is Sn = Sn-1P, the general formula for probability
of a process ending up in a certain state is: Sn = S0Pn.
In 50 days: S50=S0P50=
[0.55¿0.15¿0.30 ]
[1 0 0 ] . [0.35¿0.45¿0.20 ]^50 = [0.4207 0.3103 0.2690 ]
[0.30¿0.40¿ 0.30]
In 100 days: S100=S0P100=
[0.55¿0.15¿0.30 ]
[1 0 0 ] . [0.35¿0.45¿0.20 ] ^100= [0.4207 0.3103 0.2690 ]
[0.30¿0.40¿ 0.30]
From this we can conclude that when n → ∞, the probabilities will converge to a steady state,
indicating that in the long-term, 42.07% of weather will be sunny day, 26.9% of weather will be
cloudy day and 31.03% of weather will be rainy day. What we can see that the steady-state
probabilities of this Markov chain do not depend upon the initial state.
3. Explaining simulating weather forecast code using Markov chain:

Step 1: Creating transition matrix, which represents the probabilities of moving from one state
(e.g., sunny, rainy, cloudy) to another, based on given datas.
Step 2: Creating a list of possible weather states, which is “Sunny”, “Rainny”, “Cloudy”. Then,
we set the “intial_state” value, which is the beginning state, as 1 (Sunny).
Step 3: Setting the number of days that the programme has to forecast (in this case it is 7 days)
to the “num_days” value. After that, we initialize the “forecast” array to store the weather
predictions in “num_days”. The first day’s weather has to set to the “initial_state”.
Step 4: This loop generates the weather forecast for the remaining days:
“forecast(i-1)” finds the current day's weather.
“probabilities = transition_matrix(forecast(i-1), :)” gets the row from the transition matrix
corresponding to the current state.
“find(rand <= cumsum(probabilities), 1)” uses a random number (rand) to determine the next
day's weather based on the cumulative sum of transition probabilities.
 “rand” generates a random number between 0 and 1.
 “cumsum(probabilities)” creates a cumulative probability distribution.
 “find()” selects the next state based on this random number.
Step 5: Using the loop to print the weather predictions in “num_days” from “forecast” array.
4. References:
[1] https://iopscience.iop.org/article/10.1088/1742-6596/1848/1/012061

You might also like