Skip to content

ETHZ

Lectures

DONE Projects [3/3]

  • [x] Semester paper in economics : Public Employment

    • [x] Write e-mail.
    • [x] Decide subject and write about it.
    • [x] Read papers
    • [x] Do the regression

  • [x] Semester paper in statistics: Missing data

    • [x] Read introduction missing data.
    • [x] Play with mi.
    • [x] Use Digital Ocean API in R.

  • [x] Master thesis

    • [x] Webapp to teach math in clojure and middleman yaml

DONE Causality [2/2]

I had 5.75 (kind of deserved).

  • PDAG: Partially means no directed cycle whereas DAG means no cycle at all. CPDAG (Markov Equivalent Graphs): for every directed node, all graph have the same edge. For undirected edges, there exist at least two graphs with opposite direction of the edge. Markov equivalent if they have the same skeleton and immoralities (v-struct).
  • D-separation → Blocked Reverse engineer the definition by thinking about the independence
  • Markov blanket M: for a target node Y, it is smallest set M of nodes in the graphs that make Y independent of all the other nodes given M. In DAG: for X it contains its parents, its children and the parents of children.
  • Causal minimality: PX is not markov to any proper subgraph of 𝒢 if and only if there are no useless edge.
  • Markov: conditional d-sep ⇒ independence, faithfulness is the converse. Local: X is independent of any non descendent given its parnets.
  • Adjustment formula (tautology): the conditional probability given the parents remains the same under the interventions.
  • The goal adjustment set: evaluate interventions or causality with observational data.
  • Valid adjustment set Z: p𝒮, do(X=x)(y) = ∑z ∈ Zp𝒮(y|X=x,z)p(z) for y not in the parent of x and x, y ∉ Z.
  • Backdoor criterion: S blocks all paths from X to Y that come into X (backward path) and S contains no descendant of X.

DONE QRM [3/3]

\<2015-08-18 Die> Know the main theorems by heart: Fisher-Tippet, Pickands-Balkema-de Hann, Sklar. 5.5/6

  • [x] Read slides
  • [x] Learn slides by heart
  • [x] Do past exams

DONE Multivariate Statistics [2/2]

I had 4.5/6 (which was not deserved at all).

  • [x] Do exercises

    • [x] Serie 1
    • [x] Serie 2
    • [x] Serie 3
    • [x] Serie 4
    • [x] Serie 5

  • [x] Read scripts

DONE Computational Statistics, finish the series [2/2]

\<2015-08-18 Die> I got 6 (was a little bit easy, thanks to emacs and R-help).

  • [x] Do Series
  • [x] Read script

DONE Economics Cycle [2/2]

\<2015-06-09 Die> I had 4.75/6 at the exam.

  • [x] Read scripts
  • [x] Do Exercise

See also (generated)