spotmarket on Github

initial commit

I published the initial code to GitHub with install directions for Ubuntu. The project so far is a Python script that consumes map API jump and kill activity and inserts them into a PostgreSQL database. Details as to why I picked these two items as starting points to come later.

https://github.com/brentnowak/spotmarket

So far this has been a fun learning project; I’m getting more familiar with Python datatypes/iterators/quirks, have been leveraging arrow for any datetime work, am getting more comfortable with pandas, and have a handle on the basic differences between MySQL and PostgreSQL.

Growing Laundry List

  • Frontend – The previous project was written in a PHP framework and I have some basic JavaScript under my belt, but I am historically more of a backend/server focused person so web presentation is a weakness of mine. Django has been mentioned to me so perhaps I will do some research.
  • Implement more efficient UPSERT to improve database insertion efficiency.
  • Figure out what CREST market data to record.
  • Start working on putting the pieces together for the first analysis — more on that later.

Screenshots

Jita Jumps

2016-01-21_backup_instance_jita

Presenting data with Python Pandas.

2015-01-14_venal_ratting_df

Manually extracting the data to a csv and charting it in D3.js for now.

2016-01-18_npc_graph_1


spotmarket Prototyping

The past few months have been a series of shifting goals and commitments for me, both personally and professionally. I’ve spent more time outside running trails in the Marin Headlands and I changed jobs, landing a engineering role at a trading firm.

The winter months in San Francisco mean rain, and as such, I have been spending a lot of time reading Python tutorials, collaborating with James and Greg at element43, and working on a list of design requirements for the spotmarket trading platform.

The biggest addition to the project has been the buildout of a proper homelab server running ESXi. I’ve wanted to build out a homelab since I’m a network engineer and often tinker, but I’ve never really had a driver for the project. For the odd virtual machine or instance of a virtual switch/router that I’ve needed, I’ve leveraged an install of Oracle VirtualBox on my desktop. Now that I want to run development, production, and other testing machines, I built out a server.

Server ‘esx-01’, Uninspired Name

I ended up running ESXi 5.5 because I struggled to get my ‘unsupported’ NIC to work in 6.0. I followed a few guides to load drivers into the installation ISO, but was unable to get the setup procedure to complete properly for 6.0. My needs are basic on the VMware end so I’m sticking with 5.5 as I was able to get that to install with Realtek 8168 onboard NIC.

Details for the build:

  • ASUS M5A99FX PRO R2.0 Motherboard
  • AMD FX-8320 Vishera 8-Core 3.5 GHz CPU
  • SanDisk Ultra II 2.5″ 480GB SSD
  • G.SKILL Ripjaws X Series 16GB (2 x 8GB)

IMG_0337 IMG_0338

2016-01-09_esx_cpu

Python Learning

I was in a computer engineering program for two years in college before switching over and finishing in a telecommunications program. During that time I was exposed to assembly, C, and C++. I spent many late nights struggling with those languages, so I was apprehensive to dive back into programming.

I picked Python as the main language for this project and so far it has been outstanding. I’m getting more comfortable with the Python datatypes, iterators, and reliance on indentation. For data manipulation I’m using pandas, which for anyone that loves rows and columned data, it is second nature.

IMG_0416

Another item that has taken a lot of brain time has been the switch from MySQL to PostgreSQL. I’ve had to learn about the use of schemas, differences in data types, how to enable TCP access, and migrate my administration/query generator tool over to pgAdmin.

The Chats

If you have any comments or want to chat with me, find me on the new #k192space channel at tweetfleet.slack.com. Upcoming posts will delve into coding and talk about the reporting that I’ve started to build.


spotmarket Project Announcement

If you have followed the musings of this blog over the past two or three years, you are probably familiar with our Wallet Manger project. Growing out of the limitations of comparing market price exports from the client in Excel, James and I started a homebrew project in a framework called Yii to help us manage our market trades and have more visibility on item performance.

It wasn’t polished, lacked security features, had many missing features, and was more of a proof of concept than a properly managed project; it was a path to flex our IT muscles and learn XML, JSON, PHP, MySQL, database Indexes, and version control.

2015 ended on a high note for me as I landed a position at a quantitative trading firm, with the challenges of coding the Wallet Manager and stories about working with Eve 3rd party developers broached during the interview process.

2016 will be a year of rekindled interest in working with Eve data again for both me and my partner James. Want to refresh our homebrew project using an updated set of tools and change the focus.

There are many projects out there that record historical data from the API and present it to you. We want to go a step further and build an analysis platform that can incorporate data from the API and use it to create relationships to fuel predictions. We’re currently picking our tools and formulating requirements so be sure to watch for updates as we develop our platform.

If you want to join the discussion, head on over to #k162space on tweetfleet.slack.com.