Using the COINS data to recreate PESA reports

Jun 12, 2010 3 comments

On June 4th, 2010 HM Government released Combined Online Information System (COINS) database, the main database used by HM Treasury budgeting. I’ve written a brief overview of the data here. This post is about a little programming project I set myself to actually do something with this data.

The COINS data is in a raw format and needs to be converted to a more useful format before any useful reports can be generated. Mainly out of curiosity, I set myself a project to do something useful with this data. I wanted to do three things:

  • Extract useful subsets of the data that were small enough to be loaded into Excel.
  • Load the COINS data into an SQL database.
  • Recreate some of the government PESA(Public Expenditure Statistical Analyses) reports from the COINS data and check that my results are the same.

The project to do all this is written in python, and I’ve placed it on github here

Extracting useful subsets of the COINS data

The COINS fact_table_extract_2008_09.txt file contains 2,043,129 records that have a non-zero value field: of these 20,566 are outturn records, 1,610,900 are forecast records, 13,733 are plans records, and 397,930 are snapshot records. Older Excel spreadsheets are limited to 65535 rows, and I wanted to produce a useful data set that fit within this limit. The non-zero outturn records easily fit, I’ve made them available on rapidshare: coins2008_09 and coins2009_10.

Loading the COINS data into an SQL database.

I’ve extracted the 2008-09 and the 2009-10 COINS fact data into two SQLite databases: coins_2008_09_sqlite and coins_2009_10_sqlite

Checking my reports against published government data

HM Treasury produces Public Expenditure Statistical Analyses(PESA) reports. I wanted to recreate some of these reports from the COINS data.

I’ve produced outturn Resource, Capital, DEL and AME reports by departmental group from the 2008-09 COINS data, and checked it against the estimated 2008-09 outturn from the PESA 2009 report and the actual 2008-09 outturn from the PESA 2010 report.

The PESA data used was:

The spreadsheet PESA 2010, Chapter 1 tables which contains tables of DEL, AME and TME grouped by Resource and Capital expenditure (from webpage PESA 2010 section 1 – Budgets). 2008-09 subtotals by departmental group are provided in Column G (2008-09 outturn) of the worksheets:

  • Table 1.3 Resource budgets, 2004-05 to 2008-09 (by departmental group)
  • Table 1.6 Capital budgets, 2004-05 to 2008-09 (by departmental group)

The spreadsheet PESA 2010, Chapter 1 tables which contains tables of DEL, AME and TME grouped by Resource and Capital expenditure (from webpage Public Expenditure Statistical Analyses 2009). 2008-09 subtotals by departmental group are provided in Column G (2008-09 estimated outturn) of the worksheets:

  • Table 1.5 Resource budgets, 2003-04 to 2010-10
  • Table 1.10 Capital budgets, 2003-04 to 2010-11

I produced my report by running the python program coinssqlpesa.py against my generated 2008-09 SQLite database. (See my readme file for a full description of how to recreate these reports.)

The comparisons

The following tables give my results and the PESA result side by side for comparison.

Resource DEL by departmental group (millions)
COINS
2008-09
outturn
PESA
2010
outturn
PESA
2009
estimate
Children, Schools and Families 46,848 46,848 46,848
Health 118,899 90,278 92,455
Transport 6,245 5,083 6,546
Innovation, Universities and Skills N/A N/A 16,666
CLG Communities and Local Government 28,755 4,105
24,651
4,274
24,647
Home Office 9,198 9,198 8,926
Justice 9,234 9,235 9,283
Law Officers’ Departments 723 724 733
Defence 29,790 32,620 37,889
Foreign and Commonwealth Office 2,027 2,027 2,025
International Development 4,758 4,758 4,835
Energy and Climate Change 292 293 1,016
Business, Enterprise and Regularity Reform 175 N/A 1,594
Environment, Food and Rural Affairs 2,449 2,446 2,654
Culture, Media and Sport 1,455 1,456 1,633
Work and Pensions 7,937 7,937 8,059
Scotland 24,090 24,090 24,599
Wales 12,775 12,799 12,970
Northern Ireland Executive 7,926 7,926 8,117
Northern Ireland Office 1,177 1,177 1,342
Chancellor’s Departments 4,463 4,473 4,826
Cabinet Office 413 1,995 2,049
Independent Bodies 586 791 806
Resource AME by departmental group (millions)
COINS
2008-09
outturn
PESA
2010
outturn
PESA
2009
estimate
Children, Schools and Families -2 10,652 10,914
Health 2,870 14,984 13,934
Transport 602 603 3,878
Innovation, Universities and Skills N/A N/A 274
CLG Communities and Local Government 1,282 621
661
152
660
Home Office 709 710 363
Justice 307 439 725
Law Officers’ Departments 9 9 N/A
Defence 392 6,193 6,864
Foreign and Commonwealth Office -28 -28 -7
International Development 145 213 397
Energy and Climate Change 2,403 2,403 4,589
Business, Enterprise and Regularity Reform 1 N/A 819
Environment, Food and Rural Affairs 0 0 -52
Culture, Media and Sport 3,890 3,890 3,911
Work and Pensions 135,344 135,344 135,546
Scotland 4,456 2,495 3,129
Wales 110 138 520
Northern Ireland Executive 6,711 6,104 8,142
Northern Ireland Office 395 396 264
Chancellor’s Departments 71,836 71,209 80,757
Cabinet Office 1 7,174 7,221
Independent Bodies 9 15 24
Capital DEL by departmental group (millions)
COINS
2008-09
outturn
PESA
2010
outturn
PESA
2009
estimate
Children, Schools and Families 5,519 5,519 5,634
Health 11,602 4,370 4,561
Transport 7,182 7,252 7,283
Innovation, Universities and Skills 0 N/A 2,123
CLG Communities and Local Government 7,233 7,112
122
7,125
129
Home Office 836 836 862
Justice 912 912 975
Law Officers’ Departments 8 9 12
Defence 7,754 8,980 8,604
Foreign and Commonwealth Office 226 227 218
International Development 875 875 891
Energy and Climate Change 1,667 1,667 1,688
Business, Enterprise and Regularity Reform 2 N/A 16
Environment, Food and Rural Affairs 609 610 618
Culture, Media and Sport 841 842 791
Work and Pensions 91 91 57
Scotland 3,332 3,333 3,337
Wales 1,626 1,627 1,656
Northern Ireland Executive 1,233 1,233 1,137
Northern Ireland Office 67 68 78
Chancellor’s Departments 282 282 293
Cabinet Office 41 397 416
Independent Bodies 32 37 42
Capital AME by departmental group (millions)
COINS
2008-09
outturn
PESA
2010
outturn
PESA
2009
estimate
Children, Schools and Families 0 N/A N/A
Health 3,780 14 14
Transport 22 N/A N/A
Innovation, Universities and Skills 0 N/A 4,230
CLG Communities and Local Government 3,766 516 -
Home Office 0 N/A N/A
Justice 27 N/A N/A
Law Officers’ Departments 0 N/A N/A
Defence 85 N/A -
Foreign and Commonwealth Office 41 N/A 0
International Development 86 N/A N/A
Energy and Climate Change -35 -279 -419
Business, Enterprise and Regularity Reform -210 N/A N/A
Environment, Food and Rural Affairs -12 1 0
Culture, Media and Sport 571 572 717
Work and Pensions 126 136 142
Scotland 900 180 187
Wales 293 168 177
Northern Ireland Executive 611 378 413
Northern Ireland Office 0 N/A N/A
Chancellor’s Departments 86,001 85,822 85,590
Cabinet Office 0 N/A N/A
Independent Bodies 0 - -

The differences

There is good agreement between some of my calculated results and the PESA reports, but there are also some significant differences between the figures.

In the “Resource DEL” and “Capital DEL” tables there is generally good agreement, except for Health, Transport, Defence, Cabinet Office and Independent Bodies departmental groups. The Cabinet Office and Independent Bodies discrepancies could be because some departmental data is missing from the COINS data (see next section). I think the differences in the values for Health are probably because the PESA reports do not include the figures for Scotland, but I haven’t checked this. I have no explanation for the differences between the figures for Transport and Defence.

The agreement for the “Resource AME” and “Capital AME” tables is less good. Some of this disagreement is for the same reasons as the “DEL” differences, but I suspect the main reason for the differences is that my SQL query for the “AME” figures is not correct. (Any corrections to my SQL queries are most appreciated, as are any bug reports or corrections to my programs.)

Missing COINS departmental data

The COINS data does not seem to have spending figures for the following government departments:

  • Assets Recovery Agency
  • Department for Business, Enterprise and Regulatory Reform
  • Office of Communications
  • Forrestry Commision
  • National Investment and Loans Office
  • Royal Mint
  • Office of Government Commerce
  • Security and Intelligence Agencies

This may because the departments have been renamed, or subsumed into other departments.

Next steps

I’ve satisfied my curiosity about the COINS data. I’ve managed to recreate some of the PESA reports, and although there are differences between my reports and the PESA reports, the differences are mostly explainable. I’m not Cliff Stoll, and won’t be tracking down the exact nature of the differences. I don’t plan to do any further work on the COINS data, but may look at future data that is released, especially if it is more detailed.

Feel free to use my programs if you wish – it would nice if you credit me if you do so. I’m also interested in other programming projects that use the COINS data – feel free to post a comment with a link if you do anything interesting with the COINS data.

Looking for a small, light, quality camera?

May 14, 2010 6 comments

The success of the Olympus and Panasonic mirrorless interchangeable lens cameras shows that there is a demand for this type of camera – a small, light quality camera. Something smaller and lighter than a DSLR, but with better image quality and flexibility than a compact camera.

I’ve argued before that there is a proportion of this market segment for whom size and weight is a priority. It seems that Sony gets this: its recently announced NEX cameras demonstrate what can be done when size and weight are important design criteria.

But it’s not the weight of the size and weight of the camera that is important. It is the weight of the camera/lens/battery/memory card package that counts. Micro Four Thirds cameras have an inherent advantage here, since, for a given image quality, lenses for a m4/3 sensor can be made smaller and lighter than lenses for a APS-C image sensor.

This blog post collects together camera and lens sizes and weights. If camera size and weight is an important consideration for you, then these tables may be of some help. Note I make no attempt to compare the quality or ergonomics of any of these cameras, there are plenty of camera review sites that do that.

Mirrorless cameras

The following table gives the sizes of various mirrorless cameras, and their weights with various lenses. The pancake lens is the manufacturer’s wide angle prime lens. The zoom lens is the manufacture’s nearest equivalent to a 28-84mm (full frame) lens. The superzoom is the manufacture’s nearest equivalent to a 28-300mm (full frame) lens. All weights include batteries.

Model Weight Weight
(pancake)
Weight
(zoom)
Weight
(superzoom)
Dimensions LCD Dots
Olympus E-P1 355g 426g 505g 645g 121 x 70 x 36 mm 230,000
Olympus E-P2 355g 426g 505g 645g 121 x 70 x 36 mm 230,000
Olympus E-PL1 334g 406g 484g 624g 115 x 72 x 42 mm 230,000
Panasonic GF1 315g 415g 480g 775g 119 x 71 x 36 mm 460,000
Sony NEX-3 297g 371g 511g 821g 117 × 62 × 33 mm 920,000
Sony NEX-5 287g 361g 501g 811g 111 × 59 × 38 mm 920,000
Samsung NX10 414g 499g 612g 831g 123 × 87× 40 mm 614,000

The overall lightest camera/lens combination is the Sony NEX-5 with 16mm lens. This is thanks to the low weight of the Sony camera. With zoom lenses, the inherent size/weight advantage of micro4/3 comes into play: the Panasonic GF1 provides the lightest camera with standard zoom combination and the Olympus E-PL1 provides the lightest camera with superzoom combination. (Of course slightly lighter combinations could be obtained by mixing Panasonic and Olympus lenses and bodies.)

Note that Samsung does not seem to have got it. The NX10 is barely smaller or lighter than some of the smallest DSLRs (see next section).

DSLRs

For comparison, here are the sizes and weight of some of the smaller and lighter DSLRs:

Model Weight Weight
(pancake)
Weight
(zoom)
Weight
(superzoom)
Dimensions LCD Dots
Canon EOS 1000D 502g N/A 702g 1097g 126 x 98 x 65 mm 230,000
Nikon D60 522g N/A 787g 1082g 126 x 94 x 64 mm 230,000
Olympus E-450 426g 521g 616g N/A 130 x 91 x 53 mm 230,000

Compact cameras

And for further comparison, here are the weights and dimensions of some of the higher end compact cameras:

Model Weight Dimensions LCD Dots
Canon PowerShot G11 [28-140mm] 375g 112 x 76 x 48 mm 461,000
Canon PowerShot S90 [28-105mm] 195g 100 x 58 x 31 mm 461,000
Leica X1 [35mm] 306g 124 x 60 x 32 mm 230,000
Nikon Coolpix P6000 [28-112mm] 280g 107 x 66 x 42 mm 230,000
Panasonic Lumix DMC-LX3 [24-60mm] 265g 109 x 60 x 27 mm 460,000

Lenses

For reference, here are the weights and sizes of the lenses available for mirrorless interchangeable lens cameras:

Model Weight Dimensions Filter diameter
Olympus 9-18mm[18-36mm] f4.0-4.6 155g 57 x50 mm 52mm
Olympus 14-42mm[28-84mm] f3.5-5.6 150g 62 x 44 mm 40.5mm
Olympus 14-150mm[28-300mm] f4.0-5.6 290g 64 x 83 mm 58mm
Olympus 17mm[34mm] f2.8 71g 57 x 22 mm 37 mm
Panasonic 7-14mm[14-28mm] f4.0 300g 70 x 83 mm -
Panasonic 14-42mm[28-84mm] f3.5-5.6 165g 61 x 64 mm 52 mm
Panasonic 14-45mm[28-84mm] f3.5-5.6 195g 60 x 60 mm 52 mm
Panasonic 14-140mm[28-280mm] f4.0-5.8 460g 70 x 84 mm 62 mm
Panasonic 20mm[40mm] f1.7 100g 63 x 26 mm 46 mm
Panasonic 45mm[90mm] f2.8 225g 63 x 63 mm 46 mm
Panasonic 45-200mm[90-400mm] f4.0-5.6 380g 70 x 100 mm 52 mm
Samsung 18-55mm[28-85mm] F3.5-5.6 198g 63 x 65 mm 58 mm
Samsung 30mm[46mm] f2 85g 62 x 22 mm 43 mm
Samsung 50-200mm[77-308mm] F4-5.6 417g 70 x 101 mm 52 mm
Sony 16mm[24mm] f2.8 74g 62 x 23 mm 49 mm
Sony 18-55[27-83mm] f2.5-5.6 214g 62 x 60 mm 49 mm
Sony 18-200[27-300mm] f3.5-6.3 524g 76 x 99 mm 67 mm

Popular science on TV – call for a scientist’s cut

Apr 25, 2010 1 comment

Films are sometimes re-released as a director’s cut – a version that more closely reflects the director’s vision for the film, free from the commercial pressures of the studio.

Science programme makers could take a cue from filmmakers and produce two cuts: one for the general public and one for enthusiasts.

Take Wonders of the Solar System. This has all the makings of a great science programme: an enthusiastic and knowledgable presenter in Professor Brian Cox, great pictures of space and the earth, travel to interesting locations, interviews with scientific experts and fantastic computer simulations. Yet the programme is hard to watch – it’s like eating cabbage: I do it because I know I ought to, not because I want to. And it’s not just me. A friend of mine, who is also a science and technology enthusiast, said he fell asleep during the programme.

The main problem is that the programme is constrained to a 1-hour slot, and is edited to fit that slot. This means there is a lot of filler to sit through. Shots of Brian Cox walking up to a telescope, unnecessary shots of him driving in his car (and, of all forms of filler, presenters driving in their car is the most clichéd and overused), scenes that are portmanteaux of earlier scenes, and so on.

The second problem is that the programme, by necessity, targets a wide audience: since there are relatively few science programmes on television, such programmes have to target both the general public and those who have a deeper understanding of science. This means these programmes generally need to include explanations of the basics, and even when these explanations are good, they are not that interesting to people who already understand what is being explained. A good example is when Brian Cox explained the size of the solar system, by placing planets in their relative positions on a table, and then driving to the position of the Oort cloud. This was a dramatic illustration of the solar system’s size (and, incidentally, appropriate use of a shot of him driving), but not that interesting to someone who understands the vastness of space.

Both these problems could be solved by having two cuts of the programme – a standard version and a scientist’s cut. The scientist’s cut would differ from the standard cut in that it would not include the filler and the basic explanations. But it might include more detail in other areas.

Programme lengths no longer have to be totally dictated by the TV schedule. Perhaps the main showing needs to fit a one-hour slot, but repeats on BBC iPlayer certainly do not. There is no reason why two cuts of “Wonders of the Solar System” should not be available on iPlayer – the full version and the scientist’s cut. The programme is repeated on BBC1 and BBC4, often in the middle of the night – there is no reason why one of these repeats should not be a scientist’s cut. The programme is to be sold on DVD – there is no reason why there should not be a DVD scientist’s cut.

Despite the fact that it contains some wonderful material, I won’t be buying the DVD of Wonders of the Solar System. But if Professor Brian Cox was given full editorial control and was allowed to make a scientist’s cut of the programme, I would buy that.

Film enthusiasts get a special cut, science enthusiasts should get one too.

Eyjafjallajökull – a lesson in Icelandic and geology

Apr 17, 2010 1 comment

jökull is Icelandic for glacier. Glaciers do not erupt, volcanoes do.

The volcano underneath the glacier is called Eyjafjall. Note it’s Eyjafjall, not Eyjafjalla. Eyjafjalla is the genitive (possessive) plural form: Eyjafjallajökull literally means “eyja mountains’ (sic) glacier”. (Fjall is Icelandic for “mountain”, it is derived from the Old Norse, as is our English word “fell”. Eyja is Icelandic for “island”, it is derived from the Old Norse íeg as is our word “island” – which is why the “s” is not pronounced. “Ey” lives on in English placenames, such as Anglesey, Guernsey, Jersey and Sheppey.)

Talking about the Eyjafjallajökull eruption is not only inaccurate, it is also long-winded. Let’s be accurate and more concise and say “Eyjafjall”.

300th anniversary of copyright

Apr 10, 2010 2 comments

The Statute of Anne, the first copyright law, entered into force 300 years ago today, on the 10th April 1710. It established a copyright term of 14 years, which could be renewed for another 14 years if author was still alive when the first term expired (books already in print were granted a copyright term of 21 years).

The Copyright Act 1814 extended the copyright length to 28 years, and if the author was still alive when the term expired, the right of publication could be extended for the rest of the author’s life.

The Copyright Act 1842 extended the copyright term to the life of the author plus 7 years, or 42 years from the first date of publication, whichever was longer. Posthumously published works were provided with a 42 year copyright term.

The Copyright Act 1911 extended the term of copyright to life and fifty years (with certain exceptions).

The Copyright, Designs and Patents Act 1988 extended the copyright term to seventy years from the death of the author for literary, dramatic, musical or artistic works. If the author is unknown, copyright expires seventy years after the work is first made available to the public. If the work is computer-generated, copyright expires fifty years after the work is made.

In most countries around the world, copyright length is life of the author plus 50 years or life of the author plus 70 years

The Myth of Panic

Apr 8, 2010 Leave a comment

We see it time and again in the movies. A disaster is about to happen: a nuclear bomb is about to explode in a major city, a deadly virus has been released from a laboratory, a giant meteor is about to strike the earth, an alien spaceship is spotted heading towards the earth. And the reaction from the president is always the same: “We must not tell the public or there will be mass panic and hysteria”.

Except there won’t.

The assumption of public panic is a useful plot device for disaster movies, but as an instrument for policy formulation it is defective. Policy makers seem to assume that the public will panic, but research shows the opposite. Baruch Fischhoff, professor of social and decision sciences at Carnegie Mellon University and president of the Society for Risk Analysis, says “people, however stressed, almost always keep their wits and elevate their humanity”. In his article A Hero in Every Aisle Seat he says:

Studies of civilians’ intense experiences in the London Blitz; the cities of Japan and Germany in World War II; the 1947 smallpox outbreak in New York; the earthquake in Kobe, Japan, in 1995; and even fires have found that people, however stressed, almost always keep their wits and elevate their humanity.

In his article Panic: myth or reality, Lee Clarke, Associate Professor, Department of Sociology, Rutgers University states:

Before, during and after disasters, the ‘general public’ warrants trust and respect. Panic is often used as a justification by high-level decision makers to deny knowledge and access to the public, on the presumption that people cannot handle bad news. Research on how people respond to life threatening disasters and the stories form the World Trade Center show that people handle even the most terrifying news civilly and cooperatively. Our leaders would do well to see us as partners in recovery rather than a ‘constituency’ to be handled.

In The Swine Flu Panic That Wasn’t. Mass hysteria fails to materialize. Again. Jesse Walker says:

People are sharing information, they’re seeking out information, they’re asking questions about whether or not they have the symptoms,” says Jeannette Sutton, a researcher at the Natural Hazards Center at the University of Colorado at Boulder. “Those are not incidents of panic or hysteria. That’s rational thinking, where people are asking questions and trying to make decisions based on the information they have available to them.”

It’s not as though there haven’t been any destructive overreactions to the H1N1 flu. It’s just that they’ve come from officials, not the general public.

A John Hopkins University study dispels panic myth and suggests ways to involve the public in response to a bioterrorist attack.

Planners and policy makers have long discounted the public’s ability to participate in a response to bioterrorism, because of a belief that an attack would create mass panic and social disorder.

However, researchers at the Johns Hopkins Bloomberg School of Public Health, who reviewed the public’s response to the terrorist attacks of September 11th, the recent anthrax mailings, and other disasters concluded that the public does not react with panic but with effective and adaptive action and can be an valuable response force and that should be considered in biodefense planning.

It is a myth that a community’s first response to a crisis is panic.

The public are much smarter than portrayed – we pay much more attention to the actual reported numbers than the apocalyptic predictions of the media and politicians.

The assumption of public panic results in poor policy decisions. It’s not the public who panics, it’s the politicians.

The form of jury-less criminal trials

Apr 1, 2010 Leave a comment

The first major criminal trial to be held without a jury in Britain for more than 350 years (the Heathrow Airport Robbery Trial) concluded yesterday. Until this case, and since the abolition of the Court of Star Chamber in 1641, all serious criminal offences on indictment were tried by jury. Trial by jury traces its roots back to Article 39 of the Magna Carta signed by King John in 1215.

Trials on indictment without a jury were made possible by The Criminal Justice Act 2003. A judge may now order a jury-less trial in the specific cases of complex fraud and jury tampering.

There has been much discussion about the ramifications of the removal of the right to trial by jury in criminal trials. I think that the right to trial by jury of ones peers is an important right, and I think it was wrong to remove that right.

But, in the context of our situation where we have had our right to trial by jury removed, I want to talk about something else – the form of those jury-less trials.

The jury should be replaced, not eliminated

In a jury trial, broadly speaking, the judge rules on matters of law and is responsible for sentencing; the jury decides matters of fact (by evaluating the evidence). What’s more the jury consists of individuals with varying opinions and backgrounds – the debate that occurs between jurors in the jury room is an essential part of establishing a verdict. A judge sitting on their own has nobody to challenge their assumptions.

When a case is tried by a single judge, both the form of the trial and the dynamic of the courtroom is radically changed.

Part 7 (Trials on indictment without a jury) of The Criminal Justice Act 2003 addresses the problems of complex and lengthy trials and the problems of jury tampering. These are problems resulting from having citizen jurors, not problems resulting from a having a jury per se.

Rather than solving the problems by complete removal of the jury, I believe that the less drastic measure of replacing the citizen jury by an appointed jury should be adopted. In particular:

  1. The jury should be replaced, not combined with the role of judge. That is there should be a judge who acts as judge in the case, and a separate judge who sits unrobed in the jury box and acts as proxy for the jury. Being unrobed is important, since it is a constant reminder to the court that the judge is acting as jury, not as judge and jury.
  2. Having said (2), and mentioned the importance of debate within the jury, I think that if the jury is replaced, then it should actually be replaced by at least two judges who sit unrobed in the jury box.
  3. The sitting judge should treat the appointed jury of judges just like a citizen jury.
  4. Having three judges for a trial is undoubtedly expensive, but the expense serves as a disincentive to removing a citizen jury.

Links

Judge with no jury convicts Heathrow robbery gang

Armed raiders jailed after trial without jury

Heathrow robbers guilty: the trial without jury that made history

Four jailed for £1.75m Heathrow robbery

Law Review: No angry men

No angry men: first trial without jury begins

Tags: , , ,
Follow

Get every new post delivered to your Inbox.