S2 sampling

Original post by Jooooshy

He's modelling the mean. What method do you suggest?

The normal distribution N(3.5, 35/(12*70)) is being used to estimate the probability. It's a continuous distribution.
A dice throw is discrete.

When you model a discrete variable with a continuous distribution you need to apply a continuity correction.... :smile:

Reply 8

Jooooshy

17

Original post by Ranibizumab

The normal distribution N(3.5, 35/(12*70)) is being used to estimate the probability. It's a continuous distribution.
A dice throw is discrete.

When you model a discrete variable with a continuous distribution you need to apply a continuity correction.... :smile:

Ah forgive me, I didn't know cont. corrections came up in S2! I only did them in S3 :smile:

Reply 9

Original post by Jooooshy

Ah forgive me, I didn't know cont. corrections came up in S2! I only did them in S3 :smile:

Lol no problem. Let me know if you need anymore help :smile:

Reply 10

OP

Original post by Ranibizumab

The normal distribution N(3.5, 35/(12*70)) is being used to estimate the probability. It's a continuous distribution.
A dice throw is discrete.

When you model a discrete variable with a continuous distribution you need to apply a continuity correction.... :smile:

Uses correction but still got it wrong. What went wrong

Edit: got it :smile:

how do you know whether to add or minus 1/2n?
Posted from TSR Mobile

(edited 10 years ago)

1388868015722.jpg35.0KB

Reply 11

Congrats! I'll try and draw an explanation on how you know how to add/subtract, hang on....

Reply 12

Ok I've tried to draw an explanation. This is how I have it clear in my head at least..... Does it make sense?

Reply 13

OP

Original post by Ranibizumab

Ok I've tried to draw an explanation. This is how I have it clear in my head at least..... Does it make sense?

Thanks dude I appreciate it :smile:

Could you help me with this one please. 7a+b

If I understand 7a then I'll should be able to do 7b I think

Posted from TSR Mobile

1388870490278.jpg57.7KB

1388870598255.jpg39.7KB

Reply 14

Original post by The_Blade

Thanks dude I appreciate it :smile:

Could you help me with this one please. 7a+b

If I understand 7a then I'll should be able to do 7b I think

Posted from TSR Mobile

Is it possible to get the image the right way up? My iPad keeps rotating it upside down no matter what I do! Also if you could give me the answer I can double check I'm right before I give you any info :smile:

Edit: I've sorted the image now, just need the answer to double check :smile:

Ok no worries about the answer now I have the same book here. Right, hints coming your way shortly....

(edited 10 years ago)

Reply 15

OP

Original post by Ranibizumab

Is it possible to get the image the right way up? My iPad keeps rotating it upside down no matter what I do! Also if you could give me the answer I can double check I'm right before I give you any info :smile:

Edit: I've sorted the image now, just need the answer to double check :smile:

Ok no worries about the answer now I have the same book here. Right, hints coming your way shortly....

Was about to take another photo aha. OK man :smile:

Posted from TSR Mobile

Reply 16

Original post by The_Blade

If I understand 7a then I'll should be able to do 7b I think

Posted from TSR Mobile

Right, ok first of all you can't use a normal distribution to approximate a poison distribution UNLESS lambda is greater than 15 (2.25 is far too low).

However, there is something you can do to make lambda higher than 15. At which point you can use a normal approximation.
Remember that if you're approximating a poison distribution, the mean must equal the variance. Therefore you can't have N(lambda, lambda/n). It must be N(lambda, lambda).

...see how you get on from there, but I can give more hints if needed :smile:

Edit: lol just realised what poisson gets autocorrected to! I can spell, honest :tongue:

(edited 10 years ago)

Reply 17

OP

Original post by Ranibizumab

Right, ok first of all you can't use a normal distribution to approximate a poison distribution UNLESS lambda is greater than 15 (2.25 is far too low).

However, there is something you can do to make lambda higher than 15. At which point you can use a normal approximation.
Remember that if you're approximating a poison distribution, the mean must equal the variance. Therefore you can't have N(lambda, lambda/n). It must be N(lambda, lambda).

...see how you get on from there, but I can give more hints if needed :smile:

Edit: lol just realised what poisson gets autocorrected to! I can spell, honest :tongue:

Ha didn't even realise!

And I don't think I'm getting anywhere :frown:

Posted from TSR Mobile

1388874227494.jpg36.8KB

Reply 18

Original post by The_Blade

Ha didn't even realise!

And I don't think I'm getting anywhere :frown:

Posted from TSR Mobile

Ok no problem. Remember that Poisson distributions specify a mean for a specific time period/area etc?

You have Po~(2.25) per metre squared
They have sampled 120 metres, so to covert to per 120m^2:
Po~(2.25*120) per 120 metres squared.
....now lambda is greater than 15 and you can use a normal approximation.

Remember that you are still going to be using a continuous distribution to model a discrete distribution.
And since you have already incorporated n=120 into the distribution you don't need to divide by 120 at any point (unlike the previous question) :smile:

Reply 19