March 12th meeting
Attending: Loïc Charles, Guillaume Daudin, Béatrice Dedinger, Paul Girard, Ivan Ledezma, Ahmed Nahoudha and Pierre Hollegien.
First we welcome Ahmed Nahoudha, student in Paris-8, who will work with us for three months. He’ll be working on data from Marseille. You can read more about him here.
Guillaume Plick will start working full time from July to the end of January 2016. As a result, a first version of the database structure must be finalized by the June meeting so Guillaume P. can start working on the design of the Datascape.
Matteo could work for the Medialab for a maximum of two years as there is a position for a web developer.
A meeting is planned between Béatrice, Paul and Marie-Laure in April to discuss the SciencesPo budget.
Every partner should be able to answer for the July meeting (which needs to be scheduled) the following questions :
- What was spent?
- What are we committed to?
- Do we need to spend less or can we spend more?
Frederico d’Onofrio from Duke University and Yale University could join the team to write a paper and improve his skills on Stata.
We discussed the issues identified during previous meetings. Paul has corrected the way graphs deal with missing data. However some new issues appeared:
- When missing values, the slope of the curve is not vertical as the system identifies missing values as 0.
- Superposition of the green curve (imports) and the red curve (exports) makes it difficult to differentiate imports and exports. It could be best to have two completely separated curves.
- The vertical axis should have a log option.
- The number of «double entries» has been reduced.
- There is a discussion on world flows. Today there are three ways of obtaining world flows: current estimations, reported totals and the sum of bilateral trade.
For now we use the sum of bilateral flows in the tab “Country view” and a best guess method (recent estimations or reported totals if estimations are missing) in the tab “World view”.
We decided to implement three main changes in the structure:
- Enable the user to select in the «Country View» one of the four methods to calculate world flows (including the best guess one).
- Create a Rest of the World partner equal to the difference between the sum of bilateral flows and reported totals.
- When using the best guess method, we should implement a color system so users know for each point which method have been used (similar to one used for missing values).
- Diffusion / discussion
- In order to test it, we want to organize a test workshop.
- We are also considering a special issue on Ricardo, including some methodological discussion and first-use results.
The GitHub workflow is mostly stabilised.
In order to automate the change from the central base to sources (and vis versa), we need to create a common format. To do this a few points have been raised:
- ODT format can be used.
- The main change required concerns the variables name.
- The sorting should be the same than the one used for PDF files. It means in ascending order: Source type / year / direction / exports-imports / row number. Only for three years (1749, 1750 and 1751), the order should be: year / partner / exports-imports / row number (this has further evolved since)
- Finally some columns are left empty and were created to have a unified format or to perform a calculation (in Regional sources). Those must be erased.
We might need to find a new software to compare files as the GitHub one is not satisfying. Guillaume D. has proposed this one.
Pierre will finalize his work on the database by the end of the month, thus he will be able to work on trade of medicines. This work aims at improving our knowledge on which medicines were used at that time, how medicines usage eventually democratized (through a fall in prices and an increase in consumption), and which countries were the main trade partners. Pierre will present his first findings at the June meeting.
OFCE special issue
It is almost done. The review should be ready for April.
- Thasectra has finished transcript from Rouen, and will soon be done with Bayonne. Then Thasectra will work on Marseille on which Demba and Ahmed are already working on. Demba is also working on transcripts form Flanders (1751). Laura is checking the results.
- There is no news on data from Lyon.
- No one is available to work on colonial data. We will deal with them later.
- Loïc has found at the Académie des Sciences lots of information on data construction after the shift from old to new standards as well as on tariffs used + colonial data from 1775.
June Meeting (5th)
We discuss the June meeting and come up with the following tentative program. The aim would be to accomplish what was discussed in the December 18th 2014 meeting
- Pierre will present his work on trade of medicines
- Alain Becchia will present his work on woolens
- Béatrice will present the state of advancement of the Ricardo project
- Guillaume D., Loïc and (maybe) Pierre Gervais will
- Present the formal rules for the first-round cleaning up of the merchandise names
- Suggest «toflit-based» classifications for merchandises
- On the afternoon
- Paul will introduce a data template.
- A working session on the exploration system (by commodities, time, places) and more generally the datascape features and tools will be conducted, with an emphases on the needs of economists.
Next meeting is scheduled at 10 am, at the Médialab on April 2nd. On Mai 7th, we will discuss functionalities needed by economists. To prepare this meeting Ivan will contact other economists. The economists will probably also meet between themselves beforehand. Ivan is in charge of organizing that.