Data Transfer (Was Changing from SIMS)

[GREED]

I see your blog @GREED and I raise you:

DfE big plans for big data – Matt Smith

 

Oh I see we are playing that sort of game!!

[pcstru]

I see your blog @GREED and I raise you:

DfE big plans for big data – Matt Smith

I think that's a bit more measured but it's still somewhat overoptimistic.

 

I have been giving the issues more thought and I hope to find time to comment further. I'd quite like to flesh out some of the issues on the costs/savings and why that makes the project a political target. I'd also like to go into some of the data issues - the concepts of ownership, quality and the mismatch of a central warehouse/brokerage with functional process driven requirements. No idea when I will find the time, so maybe this thread will have some zombie resurrections to look forward to.

[GREED]

Be great to hear your thoughts. But... We on here cannot in any way influence the project. You have the opportunity to by getting involved with the dfe consultations. As I do.

 

I mean this plainly but respectfully, instead of moaning and picking holes (rightly so if there are issues from a school perspective) in a forum that cannot make any difference... and so it is just moaning, get involved and moan to the people who can make changes.

 

Also moan here too because this conversation is fascinating!

[matt40k]

I'll be interesting in your post @pcstru, are you thinking in terms of ContactPoint? IE a central data holding alot of data? My point was, it already exists, it's just automating the processes so manual files aren't used. It just doesn't seem right that flat files are used for such a critical system nowadays.

[GREED]

It's also pulling all those and more disparate siloed databases/collections together. We seem to keep missing this in our conversations.

[matt40k]

Speak for yourself, last paragraph:

It’s just this change is the DfE getting his house in order, it’s bring all the various siloed services into one roof, like the governmentis doing with all it’s websites.

[GREED]

Yer... I mean until this point. It seems to be overlooked or dismissed as a minor thing. It is a major improvement for all involved.

[CAM]

I see your blog @GREED and I raise you:

DfE big plans for big data – Matt Smith

 

The page you tried to access is classified as: New URL

 

URL : DfE big plans for big data – Matt Smith

 

Stupid filters. How new is your site?

[matt40k]

New - what, two days.

[vikpaw]

Stupid filters. How new is your site?

 

New - what, two days.

 

Don't put yourself down the 40k website has been around for ages.

I was going to read your post, then i saw the IE logo and thought, am i walking into a Sales pitch... - i know you're in bed with Microsoft

[matt40k]

Me, sell Microsoft products, I'm sure I don't know what you mean! In other news, how nice is that new Surface Pro?

 

Blog is new, website is old (by web standards anyway).

[Barcrest]

@matt40k

"It also starts to solve a lot of other issues, generating ULNs for example requires manually generating a CTF then uploading to S2S and finally manually updating your MIS"

 

Is that how you get your ULN's? I just search the pupil and if they are not there request the ULN and copy and paste it into sims. Seems like you are overcomplicating things.

[matt40k]

@Barcrest How do you request the ULN?

[GREED]

Seems like you are overcomplicating things.

@matt40k how dare YOU start complicating things for schools

[CAM]

Blog is working now. New URL is probably due to a new subdomain. I should start blogging about data but I have no professional identity to blog under.

 

Regarding ULNs, your exams officer probably obtained them one by one for the sake of BTEC exams as they needed a ULN number. Now everyone in Year 9 and above needs a ULN so they can be bulk requested from S2S and uploaded via a CTF file.

 

 

EDIT - Should note I am talking about Matt's blog getting through the filter, not mine!

Edited May 22, 2014 by CAM

[matt40k]

Ah you got my point @CAM, well done, have some points - completely worthless of course, but have them anyway.

 

My point was, data is siloed, if a organisation like Learning Records Service (LRS) needs data, it starts from scratch, it doesn't reuse data (or as well as it could). I just pointed out that reducing the extractions to all these places, we're not talking about adding more data to be extracted I might add, to 1 single extract, you could build upon that system to improve how data is returned. I admit it might not be a massive saving, but a few seconds every time you do it across thousands of schools and you start to see a benefit for very little effort of tacking something on.

 

Please do @CAM, we need more material that can be referenced. It's hard to build business cases for things when the information is dotted across a forum like EduGeek where we go off topic... ooo something shinny... not saying forums are a bad thing, debate is good, but we still need articles to be build from the discussions.

[CAM]

Hmm I can try but my sparse blog is already littered with my car and Warhammer! And my Twitter feed....

[vikpaw]

Hmm I can try but my sparse blog is already littered with my car and Warhammer! And my Twitter feed....

 

Sort your Data out Man(ager)!

[CAM]

Hey, army composition, points values and rosters count as data don't they?

[matt40k]

It only counts if you do analysis on that data, like the probability of my Land Raider destroying your jet bike squadron

[CAM]

I....have no idea how to work that out. But I just made a post about Excel if you want to read it. Link in my profile.

 

I think this just derailed again...

[matt40k]

Well, we can still see the road. We've not ran so far off that it's now a dot in the distance. Plus we need to do something whilst we wait for @pcstru to come back to us

 

Just popping back onto the road for a minute...

 

costs/savings

Remember this is done with in-house staff, if you look at the DfE spend lately they've been sending their staff on Microsoft SQL training. So cost wise it's pretty cheap. Cost savings wise, cost of training + labour of automating vs cost of doing it manual, you'll be recovering after the 2nd or 3rd time its done, also it'll be accurate - the hardest part is spec'ing and I'm assuming its already done (and to a high standard) as its already happening. This isn't going to be a big project that involves getting in some 3rd party to add-in some big name product, its in-house, ongoing development.

 

political target

Why? No extra data. No extra work for schools. Using public sector staff. Follows the one roof ideals of gov.uk that's being pushed already. This isn't ContactPoint, it might end up becoming what ContactPoint would have been, but that would be years and years and years AND years off! The main problem I saw with contactpoint was it was putting it in one place then opening it up to pretty much anyone. This is just putting it in one place (what already exists), then letting the same people access it. So rather then everyone created a data sharing agreement, you could use a template one - authorization could be like how you authorise apps on facebook (obviously not the same), ie facebook holds the data, the third party requests certain fields and you approve it.

 

concepts of ownership

Data would still be "owned" in the same way as it is now.

 

quality

Surely being more proactive would only improve the data? Surely automating would improve the quality, remove the human errors (at least from the data mangling point of view anyway)

 

Going back off the road, @CAM

Thinking about 40k thing, assuming it's a 4x4 board, it'll be something like 50:50 of getting first time (assuming it's a coin toss), then 3x dice for the heavy bolter, 50:50 of hitting, plus re-rolls as it's twin linked, then what 50:50 again for damaging, then another 2 for the lascannons, again 50:50 for hitting with re-rolls, but pretty much a sure thing on damaging, then next turn your prob move so it'll be 6:1 on hitting you, assuming that rule still applies.... yer, would LOVE to see a solution that works that out, mainly so I could just say it was wrong and get them to explain how its right lol!

[CAM]

I think I worked something out. However, you forget one very important factor. I play Dark Eldar which means I don't fight fair. Your 250pt Land Raider will be going up against two Dark Eldar Ravagers for just under the same cost at AV11 with a 5+ Invulnerable Save and mobile enough to virtually guarantee first strike. (Fast skimmers hiding behind cover!)

 

THIS USES FORMULA CALCULATIONS FROM EXCEL ROUNDED TO 2DP SO EXPECT A TON OF ROUNDING ERRORS

 

Let's take your Land Raider. It is armed with a twin-linked Heavy Bolter, two Twin-Linked Lascannons. I'll begin by working out the individual shot chance.

 

On the Heavy Bolter this is a 3+ to hit @66% further multiplied by two to represent the re-roll, 6+ to penetrate AV11 @16% and 5+ to wreck the Ravager @33%. The Ravager then has a 66% chance to fail the saving throw to completely ignore your hit. Multiply these together you get:

(0.66*2)*0.16*0.33*0.66

Multiplied by three shots this gives a 9.2% chance of scoring a penetrating hit that causes the vehicle to explode.

 

The Lascannons are 3+ to hit @66% with 2x multiplier to represent re-roll, 3+ to penetrate AV11 @66% and a 4+ to wreck the vehicle @50% followed by a 66% chance to beat the saving throw.

(0.66*2)*0.66*0.5*0.66

This gives a 28.75% chance to destroy the vehicle with one shot.

 

Then I multiply the single shot chance by the number of possible shots. This means 3*9.2% on the Heavy Bolter to make 27.6% chance of blowing up a Ravager. There are two Lascannons, so their probability will be 2*28.75% giving a 57.50% chance. Add them together to give 85.10% chance of blowing up a Ravager.

 

But, this is to blow up ONE Ravager, the other one is still alive and kicking! But you can fire one weapon at a separate target with a Land Raider. If it was the Heavy Bolter, the odds of destroying both vehicles in one turn would be 57.5%*27.60% giving 15.87% chance of two smoking wrecks. If it was a single Lascannon aimed at the other vehicle, you would be looking at:

 

(27.6+28.75)*28.75

 

Giving a 16.20% chance of two smoking wrecks. But if you focused fire over two turns you would have 85.10*85.10 giving 72.42% chance of victory!

 

 

Now what if the Ravagers reply?

 

A single Dark lance will need a 3+ to hit @66%, reduces your AV to 12 requiring just 5+ to penetrate @33% and a 6+ @16% to make a Land Raider explode, giving a chance of just 3.48% to destroy the vehicle in one shot. But each Ravager has three of these and there are two Ravagers, giving six lances in total. 3.48*6 makes 20.91% chance of destroying the Land Raider in one hit if both Ravagers focused fire. If one Ravager is destroyed and the second survives, the chance of destroying the Land Raider becomes 20.91*10.45 making 2.19% so focus firing is definitely the best way to defeat the Dark Eldar.

 

This of course doesn't take into account Glancing Hits if you don't quite make the penetration roll which further complicate matters. It also nearly midnight, I am sitting here working out the cumulative probability in a web of dice rolls, tactics and unequal situations and it is probably wrong anyway. I'll let you look over the math if only for a curiosity. And don't forget those Excel rounding errors.

 

Research sources:

Warhammer 40k Tactics: Probabilities made Easy

How to Calculate Cumulative Probability | eHow

 

Working out pad:

http://1drv.ms/1nwu6p0

Edited May 22, 2014 by CAM
Found an error already...

[GREED]

Mods please split the thread