## Exercise 3

Now we are going to use our `n`-grams to generate some fake text according to a Markov model. Here's how the Markov model of order `n` works:

### A. Compute (`n`+1)-gram occurrence frequencies

You have already done this in Exercise 2!  

### B. Pick a starting (`n`+1)-gram

The starting (`n`+1)-gram can be selected at random, or the user can specify it.

### C. Generate Text

Now we generate text one character at a time. To do so:

1. Look at the most recent `n` characters in our generated text. Say that `n = 3` and the 3 most recent character are `the`. 
2. We then look at our list of `n+1`-grams, and focus on grams whose first `n` characters match. Examples matching `the` include `them`, `the `, `thei`, and so on. 
3. We pick a random one of these `n+1`-grams, weighted according to its number of occurrences. 
4. The final character of this new `n+1` gram is our next letter.

For example, if there are 3 occurrences of `them`, 4 occurrences of `the `, and 1 occurrences of `thei` in the n-gram dictionary, then our next character is `m` with probabiliy 3/8, `[space]` with probability 1/2, and `i` with probability `1/8`.

**Remember**: the ***3rd***-order model requires you to compute ***4***-grams.

## What you should do

Write a function that generates synthetic text according to an `n`-th order Markov model. It should have the following arguments:

- `s`, the input string of real text. 
- `n`, the order of the model. 
- `length`, the size of the text to generate. Use a default value of 100. 
-  `seed`, the initial string that gets the Markov model started. I used `"Emma Woodhouse"` (the full name of the protagonist of the novel) as my `seed`, but any subset of `s` of length `n+1` or larger will work.

Demonstrate the output of your function for a couple different choices of the order `n`.

## Expected Output

Here are a few examples of the output of this function. Because of randomness, your results won't look exactly like this, but they should be qualitatively similar.

```python
markov_text(s, n = 2, length = 200, seed = "Emma Woodhouse")
```
```
Emma Woodhouse ne goo thimser. John mile sawas amintrought will on I kink you kno but every sh inat he fing as sat buty aft from the it. She cousency ined, yount; ate nambery quirld diall yethery, yould hat earatte
```
```python
markov_text(s, n = 4, length = 200, seed = "Emma Woodhouse")
```

```
Emma Woodhouse!"—Emma, as love,            Kitty, only this person no infering ever, while, and tried very were no do be very friendly and into aid,    Man's me to loudness of Harriet's. Harriet belonger opinion an
```

```python
markov_text(s, n = 10, length = 200, seed = "Emma Woodhouse")
```

```
Emma Woodhouse's party could be acceptable to them, that if she ever were disposed to think of nothing but good. It will be an excellent charade remains, fit for any acquainted with the child was given up to them.
```

## Notes and Hints

***Hint***: A good function for performing the random choice is the `choices()` function in the `random` module. You can use it like this:

```python
import random

options = ["One", "Two", "Three"]
weights = [1, 2, 3] # "Two" is twice as likely as "One", "Three" three times as likely.

random.choices(options, weights) 
```

```
['One'] # output
```

The first and second arguments must be lists of equal length. Note also that the return value is a list -- if you want the value *in* the list, you need to get it out via indexing.  

***Hint***: The first thing your function should do is call `count_ngrams` above to generate the required dictionary. Then, handle the logic described above in the main loop.

and the code I did for exercise 2 was :

def count_ngrams(s,n):
   frequency=[]
   for i in range(len(s)-(n-1)):
       frequency.append(s[i:i+n])
   freq= {}
   for i in frequency:
       freq[i] = freq.get(i,0)+1
   print(freq)
   
count_ngrams("tortoise",n=2)