Considering multidimensional data to fit into model

[xinleizhao]

with each earthquake/volcano eruption, add information such as temperature, moon phase, GPS movement and if the model can reveal some of the relationships between these factors that will be great.

[xinleizhao]

For moon phase, it can be added to every day as a prefix or postfix for each day. library astral will be used. Then we can filter out sequences that only contain moon phases since for closed sequential it is easier.

[xinleizhao]

used PyEphem library for moon phase since documentation for astral is not very clear.
reference: http://stackoverflow.com/questions/2526815/moon-lunar-phase-algorithm

moon phase data is already generated in sequences, preferably every sequence contains one month, thus every sequence is a month cycle.

still have no idea how temperature data and other data can be added. Reading papers.

[xinleizhao]

magnitudes and depths are added to the sequence

[xinleizhao]

For temperature, only monthly average global data is found. To read: http://hanj.cs.illinois.edu/pdf/fpstm03.pdf

[xinleizhao]

temperature (two digits, changes approx. every month): first digit: 4 for colder compared to previous month 5 for warmer or equal to previous month second digit: 6 for below global average temperature 7 for above or equal to global average temperature

moon phase (changes approx. every 7 to 8 days): 0 for moon phase day (date - previous new moon) [0, 8) 1 for moon phase day [8, 16) 2 for moon phase day [16, 24) 3 for moon phase day [24, 30)

new sequence will be like this: each line of sequence is a phase:

Sequence Model:

<Moon Phase 0> + <Phase Average Temperature> + Event on Day 1, Event on Day 2, ... //Sequence 1
<Moon Phase 1> + <Phase Average Temperature> + EventByDay, EventByDay, ... //Sequence 2
...

<Moon Phase 0> + <Phase Average Temperature> + Event on Day 1, Event on Day 2, ... //Sequence 5

[xinleizhao]

temp is done