Market size of public cloud computing

I am currently taking a course at Stanford University called “Leading trends in Information Technology”. In one of our previous lectures we were taught about cloud computing, which is a subject of current interest and an industry of rapid growth. I have decided to look on the current and expected market size of public cloud computing. According to the analyst firm IDC is the worldwide public cloud computing spending expected to grow from $70 billion in 2015 to more than $141 billion in 2019 [1]. This is an expected increase of 19.4% compound annual growth rate (CAGR).

However, it appears that the annual spending numbers rest on widely different methods of measurement, when comparing the numbers from the two large analyst firms, IDC and Gartner. According to Gartner, the total spending on worldwide public cloud computing reached $175 billion in 2015 and the total spending is expected to grow up to $204 billion in 2016 [2]. In addition, Gartner expects a growth rate of 16.5% from 2015 to 2016. These numbers are significantly contradicting with the numbers from IDC.

Taking a further look into the two firm’s specifications of the annually cloud category “SaaS” (only available numbers), is it hard to discover a close connection between the numbers. An illustration of the numbers for each firm appears in the table below.

The numbers above exemplify the major deviations in assessment of the market size in 2015 for the cloud type “SaaS” with a difference of $15.6 bn. An even larger difference on $120.6 billion appears when comparing the remainder of the cloud categories (other).

So which numbers is correct? I have compared the numbers above with another study on the same topic prepared by the analyst firm Market Line [4]. Market Line’s analysis estimates the cloud computing market size in 2015 to $63.7 billion and expect a CAGR on 41.3% for the next five years. Moreover, the analysis estimates the market size to grow to $118.1 billion in 2019 and $174 billion in 2020, which is found to be more consistent with the numbers from the IDC analysis.

So the question remains: what could cause the dissimilarities in the numbers presented from the different analyst firms? This is some of my rough estimates: 
1) Dissimilarities in the classification of the different cloud types 
2) Dissimilarities in the judgement of whether or not a specific service could be defined as cloud computing 
3) It is fixed on a discretionary basis how much revenue to allocate as “public cloud services” when analyzing the large public cloud providers (Amazon, Google, Microsoft etc.)
4) Gartner’s largest category “Cloud advertising” could have been omitted in the IDC analysis

Please feel free to comment/challenge the above.

[1] https://www.idc.com/getdoc.jsp?containerId=prUS40960516
[2] http://www.gartner.com/newsroom/id/3188817
[3] “Software as a Service (SaaS) will remain the dominant cloud computing type, capturing more than two thirds of all public cloud spending through most of the forecast period.” (of $70 bn.) — from the Gartner link above
[4] Market Line Industry Profile — Cloud computing in the United states, December 2015